We don’t have any more information on how the British Polling Council’s review of the election polls will progress beyond it being chaired by Pat Sturgis, but several pollsters have given some thoughts today beyond the initial “We got it wrong and we’ll look at it” statements most pollsters put out on Friday. Obviously no one comes to any conclusions yet – there’s a lot of data to go through and we need thoughtful analysis and solutions rather than jumping at the first possibility that raises its head – but they are all interesting reads:

Peter Kellner of YouGov has written an overview here (a longer version of his article in the Sunday Times at the weekend), covering some of the potential causes of error like poor sampling, late swing and “shy Tories”.

Martin Boon of ICM has written a detailed deconstruction of ICM’s final poll which would be have been an interesting piece anyway in terms of giving a great overview of how the different parts of ICM’s methodology come together to turn the raw figures into the final headline VI. Martin concludes that all of ICM’s techniques seemed to make the poll more accurate, but the sample itself seemed to be at fault (and he raises the pessimistic possibility that sampling techniques may no longer be up to delivering decent samples)

Andrew Cooper of Populus has written an article in the Guardian here – despite the headline most of the article isn’t about what Cameron should do, but about how the polls did.

Finally ComRes have an overview on their site, discussing possibilities like differential response and the need to identify likely voters more accurately.

674 Responses to “The polling post-mortem – some pollsters’ thoughts”

1 9 10 11 12 13 14

    @” should not be interpreted as Britain being ideologically Tory.”

    I think this is a mistake -I mean looking for an entity which is “ideologically Tory”-either in the electorate or in this Prime Minister.

  2. Colin

    No, they are voters that no party can rely on (not a bad thing really in a democracy).

    I have said before that I believe that in Great Britain, roughly 1/4 of the voter are on the left, 1/4 on the right, 1/4 are open to offers and 1/4 don’t care.

  3. Colin

    I know that “ideological Tory” is really a contradiction in terms traditionally. It was when it stopped being a contradiction in terms that things went wrong for them.

  4. @ Hawthorn

    If you can’t or don’t want to answer, the apologies:

    Are you saying that both the raw sample and then the correcting of the sample are skewed, inappropriate, thus increasing the probability of both false positives and false negatives, respectively? And because of some, unemphasised (though probably about affiliations and/or demographics) basic assumptions, it skewed to a particular direction?

  5. I haven’t commented before, but as a lawyer, I think there are a number of misconceptions about the British Bill of Rights that are being circulated both in the media and on this board. Hopefully this will help.

    The European Convention on Human Rights is an international treaty and the UK was a signatory for many decades before the Human Rights Act was passed. Under English law, international treaties are not considered part of domestic law. In short, you can’t sue the government or anyone else based on breach of a treaty. By the way, I understand that is the position in many jurisdictions, though some, such as Germany, are ‘monist’ and treaties have different status.

    Repealing the HRA will not affect the UK’s international obligations to respect Human Rights as set out in the ECHR. It will simply mean the domestic courts cannot adjudicate on allegations of breach of the ECHR.

    The HRA contained a provision that the domestic courts should give effect to decisions of the European Court of Human Rights which was controversial – partly because it seemed to subordinate national courts to the courts of an international body, and partly because domestic courts have a different approach to court decisions than international courts – while case law is relevant to each, English common law is based on the principle that earlier court decisions are binding on later courts. That has no equivalent in the European Court of Human Rights, which has never been bound to follow its own earlier decisions.

    One point to remember about this is the fact English Common law is based on case law is linked to but logically separate from the fact that English law has – compared to most continental jurisdictions – favoured legal certainty over judicial discretion. The two should not be conflated, and while a British Bill of Rights would respect the traditional approach to case law, it can hardly avoid broad statements of principle.

    The simple way to introduce a British Bill of Rights would be to:
    1. Repeal the HRA and return to the pre 1998 position
    2. Substitute a new Bill of Rights restating the Convention obligations but without requiring English Courts to consider the ECHR jurisprudence.

    This would mean the UK would only ever be in breach of the Convention insofar as the English courts differed in their interpretation of the Convention from the ECHR.

    Clearly, the British Bill of Rights could depart from the wording of the Convention. In that case the UK would be in breach of its treaty obligations in any case where the altered wording made a difference to someone’s rights.

    I don’t think anyone is suggesting repudiating the Convention completely, so that isn’t really the issue. Likewise, all member states have been found in breach of the Convention multiple times, so being found in breach is not a major issue. Nor need it damage public opinion of the UK government if it’s found in breach, since the public may agree with it – and frequently do!

    I’m not an expert in Scots law, but I suspect the points being made about the effect of repeal in Scotland (and for that matter NI) are essentially political, not legal. If Westminster decides to repeal the HRA, it has power to do so, even if that breaks a treaty or rewrites the devolution settlement. The real issue will therefore be how this will be perceived politically.

    It would be fascinating to see how this plays out politically and thus in polling. I suspect it will depend on how each side spins its narrative to ordinary people, who cannot be expected to understand the niceties of the legal position. There are opportunities and pitfalls for both sides in this regard.

  6. @colin – “And if the sort of money needed to achieve Messina’s results is higher than the UK industry can/will spend, then is the conclusion that far less frequent , but much more effective & informative polling is the only solution which will restore credibility?”

    Not too sure it’s all about money. The Labour reports suggested that they tried to replicate pollsters results by changing methodology but couldn’t, implying that there wasn’t anything wrong per se with relative cheap online panels. It seemed to be far more the ordering of the VI question and getting respondents into an equivalent mindset as when they vote.

    @J R Tomlin
    “I find the idea that fear of the SNP cost Labour 90+ seats fairly absurd to be frank.”

    Don’t think anyone has said that. I think it’s a near 100% certainty that it cost Labour sufficient votes to enable a Con majority, however. I really think this is unarguable to be honest, and it being backed up by data from the parties own polling plus strong evidence from public polls. (ComRes found a Lab/SNP alliance the least favourite outcome amongst undecided voters I recall).

    @Laszlo – “I don’t really buy this “parties’ internal polling was better”.”

    I think you have to. There is no point reveling in a state of denial.

    One way to look at this was to examine where the parties were putting in their effort.

    Earlier in the campaign there was an LSE report analysis where the parties were putting their money. They concluded that last year Cons had cut spending in the tightest Con/Lab marginals very heavily, and were moving their resources to Lab targets much higher up the list. The implication here was that they had written off a batch of tight marginals.

    During the campaign, and especially towards the end, it was clear they were back to working the tighter marginals very hard. Cameron visited Carlisle on the last day (target Lab swing 1.0%) and they were active in a swathe of LD target seats as well.

    Lab, by contrasts, was moving resource into those tight marginals that everyone previously thought were in the bag. A big effort in Stockton South, for example.

    I’m a firm believer in watching where the parties are focusing their campaigns as a good indicator as to how they think the battle is going. I was getting a bit nervous seeing them ramp things up in places like Twickenham, and I thought it was odd to target Carlisle with your number 1 campaign weapon, but I too was fooled by the public polls.

    In retrospect, I think it’s a very straightforward step to agree that the parties had a far more accurate picture than the public pollsters, as their actions over the final week of the campaign clearly showed where they thought there were battles they could win, and where conversely they accepted they had already lost. Denying this risks being contrarian simply for the sake of it.

  7. Laszlo

    The concern was more about the fact that the sample wasn’t truly random, so the underlying assumptions beneath the confidence interval calculations were not met. The statistician wanted to add the design factor to confidence interval.

    The observed effect in the data was very strong, so we ended up adding caveats as necessary.

    I don’t think that is why the opinion polling went wrong, but it is more an observation that polling companies do things to a price and are not up to an academic standard. I am also pointing out that the people who actually do this work are not all as well educated in this stuff as Peter Kellner or Anthony Wells.

    We also have conversations about Frequentist vs Bayesian statistics as well, which made the discussion here on that subject very informative. I should point out that I am an analyst rather than a statistician (those tasks require very different intellectual processes).

  8. Some preliminary data on the electoral mountain that Labour faces.

    Of the seats the Conservatives won from Labour on Thursday, 6 can be retaken with a swing of under 1%, and the other with a swing of less than 2%. But in each of these, the Conservatives will have the benefit of a new incumbent MP to make life difficult.

    The new Labour targets for seats held by the Conservatives split as follows, based on the swing needed to take the seats:
    <1%, 6 seats
    1% to <2%, 2 seats
    2% to <3%, 6 seats
    3% to <4%, 8 seats
    4% to <5%, 15 seats
    5% to <6%, 12 seats
    6% to <7%, 13 seats
    7% to <8%, 7 seats
    8% to <9%, 9 seats
    9% to <10%, 7 seats

    So Labour has now to climb up the sides of an unfertile valley before it can get to a point where there are increasing numbers of Conservative seats to harvest.

    The large number in the 4% to <7% range is because the Conservatives achieved a swing from Labour in most of Labour's targets on Thursday. In trying to explain the result on Thursday, we shouldn't discount the effectiveness of the Conservative operation in marginal seats, which was driven by their national rather than local campaigns.

    Once someone publishes a comprehensive spreadsheet of all the results, I won't be surprised if it turns out that with a UNS, the Conservatives would now be winning more seats than Labour on an equal share of the national vote.

    And that's before the Boundary Commission, or the full impact of IER in the next round of registration.

  9. Note: The table above is of the seats the Conservatives already held, EXCLUDING the 7 seats that they won from Labour in 2015, since the difficulty of the latter for Labour will be enhanced with new incumbency effects.

    i.e. There are 12 in the <1% range, but only 6 which the Conservatives already held.

  10. @neilj

    2005 was probably the peak of the skew in the electoral system favouring Lab. They won the popular vote 35 – 32, yet won >150 more seats than Con. IIRC the Con won a plurality of votes in England in 2005, yet Labour won an overall majority of seats. That probably wouldn’t happen now, even if there was a 3-4 point uniform swing favouring Lab in England. And certainly won’t happen if the Tories push through boundary reform.


    I think that is a very fair summary of the position. Of course all that will change if the government have a major problem with the economy as the election nears.

  12. @ Hawthorn

    Thank you. it makes sense.

  13. “Thank you. it makes sense.”


    Thank God something does…

  14. Martyn
    The result of the GE tells us that a majority of people are sick and tired of the “human rights” situation that has existed for years. Those very many, who will be delighted to see some changes made are not all fascists and jesuits as you suggest.

    The ALL TORY cabinet banging on the table. Yes, I think we should hold a referendum to make the GE result void.
    Frankly monsieur I think you and your friends have much bigger thinks to worry about.

  16. As it came up last night (at least in my mind), Hawthorn reminded me just earlier, and I had a bit of boasting promise in exchanges with Unicorn.

    So a few words about about my failure with creating a Bayesian model for VI forecasting (I studied it 30 years ago, and recently used it on a concrete business decision making model, hence was the idea).

    I had an epistemological problem, as I consider each poll as distinct entities (successfully forgetting it in the last few weeks), thus could not attribute a trend to it.

    Option 1: you draw a sub-sample from a large sample with the assumption that each parties have equal chance. You examine it, and by using Bayesian calculations, you correct your equal chance, and use the corrected ones for your next sub-sample, and so on. Under this logic, the shifts on the odds on party votes will keep on reducing, and you will need a massive change to change it. Thus, you would really find something unique, otherwise the odds would not really change. This would provide a quite scientific basis for making the assumptions for likely VIs prior to sampling. I obviously couldn’t do this without the access to the raw data.

    Option 2: You have several samples, keep them separate, and poll them day after day using the method above, while using Bayesian techniques to measure the relative changes in the samples in comparison to each other (This latter one is quite difficult methodologically, but there are several routines). And use the shifts between the sample groups to measure the change in VI. I couldn’t do it for the same reason as in Option 1.

    Option 3: give up your epistemological conviction and treat each poll as a sample of a population within a set time interval. As I didn’t have any other choice, I tested this. The result was quite astonishing. It was literally all over the place. Those small 1-2% changes were hugely amplified. I tried for different periods, tried raw data and filtered data, it was the same. I checked if I was right with the methodology, but I could not find anything wrong. So Bayes said that the VI was highly unstable (but in both directions), when everything in the tables suggested stability. It was true for all parties. At this point I gave it up (and decided to look at NowCast).

    I went back to the data on Saturday. What I did was completely invalid as I had no justifiable reason for doing what I did: I took out one party each time and ran the routine for 5 periods (I was restricted by time). The stability of the Bayesian odds became more settled (but still not showing that much, although more, Conservative leads), although still dancing. hence my comments in the last few days about LibDems. however, what I did is not valid.

    Sorry, I promised a few words and it became long …

    Sorry bigger THINGS to worry about. But big thinks also are required.

  18. Sorry, forgot to close the italics …

  19. @James
    “That probably wouldn’t happen now, even if there was a 3-4 point uniform swing favouring Lab in England. And certainly won’t happen if the Tories push through boundary reform.”

    On the figures that I’ve just given you, with a 3-4 uniform swing favouring Lab in England, together with the predictable incumbency bonus in the seats the Conservatives won from Labour:
    (a) the parties would be about level pegging in national vote share and
    (b) the Conservatives would have over 300 seats still and at least 40 more than Labour

    I can’t quote exact figures because I don’t have the data to calculate how the effect of a Con-Lab swing would play out in other seats where the LDs are challenging the Conservatives, and also we can’t be sure of the effect of the new incumbency bonus, but it’s in that sort of ball park.

  20. PS. That also applies to a 3-4% swing in GB, not just England.

  21. Is there any betting market on how long Jim Murphy can hang on?

  22. We may have our first by election in a couple of months with Zac Goldsmith making an unequivocal promise he would resign if the Heathrow expansion goes ahead. A very safe conservative seat but may be more opportunities for the pollsters to make up for their disappointing performance at the GE

  23. @The Other Howard – if you are still hovering, any chance you could let me know the result of the next election?

    Saves me a lot of time and trouble if I know what’s coming.



  24. Phil

    Assuming that Labour and the Tories do not lose/gain any seats elsewhere, Labour would need to gain 50 seats from the Tories to become the largest party. I make that around 5-6% swing based on your calculations. That is slightly more than Cameron got (~5%), but a lot less than Blair (~10%).

    Only a <1% swing is needed to have another hung parliament. That is less than William Hague achieved.

    The boundary changes (if they happen) would negate any incumbency effect. I think the overall effects would be fairly small as some have materialised anyway (you risk double counting them).

    To conclude, if Labour do not recover in Scotland, then they need to slightly better than Cameron to be a minority government, and about as well as Blair to gain an overall majority.

    If the SNP maintain the seats they do, and Scotland does not go independent, then majority government will likely remain an aberration and a strong majority government extremely difficult to achieve.

    It would easier for the Tories to gain a majority than Labour, but easier for Labour to maintain minority government. There would be a wide area where stalemate would occur unless the SNP propped up the Tories.

  25. I would add that those swings need to happen in the right places.

  26. Heathrow is one of those little issues that can cause big problems.

    If Zac did end up resigning, a 20% swing is one of those by election results that Liberals used to achieve against Tories quite often, in a by gone age. Feed in some nonsense over human rights and a Lib Dem party now not associated with the government and perhaps the local conditions might suit a Lib Dem revival.

  27. Zac Goldsmith is a popular local MP, if he does resign and stands as an independent I suppose anything could happen

  28. @Alec

    I was thinking that the Lib Dems could benefit anyway. If the Conservatives mess up, particularly over the snoopers charter and other bills (which I’m not saying they will), the Lib Dems could articulate the ‘see we were doing a lot of good in government’ line pretty strongly, which might help them pick back up votes.

  29. @neil j

    “What with the unpopular wars and charges of lying/dodgy dossier etc, I still do not know how they won the 2005 election so convincingly.”

    They didn’t actually win it convincingly, they benefited from the massive skew which was still there in 2005. I will give you voteshare and seat numbers:

    Labour – 35.2% = 355 seats
    Tories – 32.4% = 198 seats

    Compare with 2015:

    Labour – 30.4% = 232 seats
    Tories – 36.9% = 331 seats

  30. It’s even more incredible when you remember that despite Labour’s 2015 disadvantages of losing 40 seats in Scotland AND piling up useless votes in London/North East…. they’re still NOWHERE near as shafted as the Tories during the Blair years.

  31. @iuvenis

    In respect of Scotland, the problem with the proposals isn’t so much the repeal of the HRA, but the imposition of the ‘British Bill of Rights’. HRA is UK law. so Westminster can amend or repeal it as it pleases.

    The problem with the ‘British Bill of Rights’ is that it would apply to both reserved and devolved matters. The fact it applies to devolved matters would then normally mean it would require consent from the devolved institution(s) affected (Sewel convention). An SNP minister said in Holyrood yesterday that they would not give this consent.

    Hence the potential for a stand-off.


  32. @Omnishambles

    That’s why FPTP really sucks.

  33. The skew isn’t in favour of either party, it is in favour of a particular segment of the electorate.

    THAT is the real case for electoral reform. As the country (England as well as the UK), there is a risk of extreme political instability.

    Indeed, that is already the case for Scotland (I would say that countries don’t get much less stable than being close to breaking up).

  34. I note that the government are looking to amend the FoI Act because it did not like the Supreme Court’s judgment.

    Ironic indeed given the stuff on the HRA and Strasbourg.

    This government is already looking authoritarian. Perhaps the Liberals aren’t dead yet after all.

  35. Have been skimming a couple of old threads from the middle of the campaign, just to remind myself what the discourse about the SNP effect was.

    Can I just reproduce the following, from Peter Crawford, which I think was quite prescient. He didn’t spot the Tories actually winning, but did seem to spot a tightening of the race in English marginals, and (probably) correctly ascribed this to the SNP.

    It is getting close…i have been tracking the England only swings on the England only polls as shown on wiki.

    Until about 2 weeks ago, the 1 month average swing from con to labour was about 4.5-5%…in the last two weeks the average has been 3.5%. On a UNS basis this means the loss of about 42 seats from tory to labour…

    If this swing gets below 2.75% con to labour, Miliband will struggle to get the 35 seats he needs to block the tories…that’s a 0.75% swing to the tories in the last 9 days in England….

    This could happen. The vote Labour get SNP run govt. line is working I would suggest.

  36. @hawthorn

    Not dead yet, that’s true. Latest figures are 10,000 new members since the election

  37. If the government goes ahead with reducing the number of constituencies, it will be a major factor if there would be more homogeneous constituencies in sociology-economic terms, or more heterogeneous (likely creating more marginals, but also pseudo-marginals).

  38. Alec

    I have actually posted a few times since the election, including this morning. In one of them i said i would not dream of attempting to forecast the election until 2018 after the EU referendum.(assuming I’m still here).

    However what i would say is that if the Government keeps the economy moving forward, settles the Scottish question, including EVEL to the satisfaction of the rUK

  39. Alec,

    I don’t know what happened but to finish…………..

    and gets through the EU referendum without a major row with his own MP’s then I think it very hard for Labour to win in 2020.

    A lot of big “ifs” of course hence my delaying any forecast until 2018.

  40. TOH

    If the Government does well over its term, then it should win at least a very good plurality in 2020.

    That is also true of every other democratic administration ever.


    “…Martyn. The result of the GE tells us that a majority of people are sick and tired of the “human rights” situation that has existed for years. Those very many, who will be delighted to see some changes made are not all fascists and jesuits as you suggest…”

    Actually I was referring to Islamists. Jesuits have far more sense.

    But returning to my point. Ceding absolute authority to others (the concept that the state is all powerful, trustworthy and infallible: the state as parent) is not usually held to be a right-wing stance: hence my point that such a position is usually held by extreme left-wingers and religious fanatics. I agree with your point that it is popular. My point is that by adopting it you have become left-wing.

  42. As a very broad hypothesis, wouldn’t larger constituencies tend towards heterogeneity? There aren’t that many “rich areas” of the UK that don’t have a “poor area” somewhere nearby.

    Of course, the exact balance of that mix would determine whether lots of voters from one party got stranded uselessly in the other parties’ seats, but their existence should at least give some hope of seat change in “exceptional” years.

    To take my city for example, if you combined both the all-Plymouth seats together you’d get one marginal tending to Labour rather than one ultra marginal, and one relatively safe (although they lost it on May 7th) Labour seat. In a year when Labour were ahead in the polls they could rely on taking both seats, and if merged together they still could.

    However, if you merged Moor View with West Devon, you’d take a whole load of Labour council estates and bury them in the Tory rural idyll of the Tamar valley and West Dartmoor, robbing Labour of a winnable seat and still leaving West Devon a Tory seat (with LDs an outside prospect maybe, once they recover and given some ABT voting). If you merged Sutton and Devonport with South West Devon, you’d again bury some council estates in the well-heeled suburbs of Plympton and Plymstock and the Tory-leaning towns to the East of Plymouth.

    It would all depend on “how you did it”. That’s why the US has gerrymandering. My fear is that the strict political neutrality of the Boundary Commission means the party policial bias might be virtually random. Wouldn’t it be better to have some “reverse gerrymandering” by specifically trying to create balanced constituencies that would be probably swing seats?

  43. Having spoken to a few friends, one of the reasons an ex-Labour voter switched to Con was because not because of fear of the SNP, but from fear of left-wing parties in general (SNP/Green) who would try to pull Labour to the left.

    Keep hearing ‘fear of nationalists’ being a reason people voted Con, but could it instead be fear of Socialism? If SNP were more liberal, would Labour have done better? Was it fear of socialism, or nationalism?

  44. My homespun philosophy is that politics is like snooker.

    It doesn’t matter how good you are, you don’t get a chance until the other guy misses.

  45. @JStephenson,

    I’ve raised that suggestion myself. Was it the talk of “progressive alliance” that made the difference, or just the Scottishness of the SNP?

    After all, Labour has always brought a lot of Scottish influence into the governance of the UK. It didn’t seem to hurt Blair.

    If English voters shied away from being ruled by SNP MPs when they would have been happy to be ruled by LiS MPs, does that point to it being about politics/economics rather than accents?

    I’m not so sure. There is a plausible argument that most voters are not particularly cognisant of the general political spectrum, or of their own place (or the place of political parties) on it.


    I agree with you, exactly why I said what i did.

    Simples :-)

  47. JStephenson

    “Was it fear of socialism, or nationalism?”

    probably both judging by responses from friends and aquaintances.

  48. @ JStephenson @ Neil A

    I don’t know if it affected the voting more or not, but in the last two weeks EM’s line was quite crystallised around social democracy (not socialism, but it could have been lost on the voters). Labour, at least in England, completely dropped the anti-nationalist line.

    Again, considering this, and how the campaign started, the only reason I can attribute to it is that he was told that it was working (or his personal inclination, which is social democracy, was let loose (and lose …) by others in the leadership who stopped giving him any negative feedback, and let him sleepwalk. This doesn’t sound very plausible to me. And I hope it wasn’t the case.)

  49. May I make a simple but rather obvious comment?


    Last year I posted a prediction on this site which used historical precedent to predict a 7% Tory lead in May 2015. It can be read here:


    It was also posted here on the front page.

    Since then I posted a number of updated projections which all estimated Tory leads of between 1%-7%. My own final estimate (posted on 31/12/14 in the “Labour Target Seats” thread in the “Constituency Guide” section of this blog) estimated a Tory lead of 6%. When asked if I stood by this a few weeks ago on the Ealing Central & Acton thread I said I did.

    It pains me to say it turned out to be true. Very few of you agreed with the estimate and while many politely disagreed (e.g. Barnaby Marder) a couple of you here on the front page were actually quite abusive about it (e.g. Crossbat111 and James Peel).

    Any chance of an apology from either of you?

    For the record, ‘Swing back’ sometimes occurs gradually during a campaign or during the run-up to a campaign but occasionally it happens at the last minute (as in 1987, 1992 and again now in 2015).

    But it was obvious all along that it was bound to happen at some stage, and the evidence was always there in the underlying poll data relating to perceived economic competence and leadership ability.

    Swing back is a fact. I rest my case.

    (P.S.:- Just for the record there is one aspect of the election result on which I am willing to hold up my hand and admit I got it terribly wrong: I mistakenly thought Labour would arrest the SNP surge in Scotland. It really bothers me that our parliament now has 56 MPs who think the people of the East End of Glasgow are the “deserving poor” while those in the East End of London are the “undeserving poor”).

  50. JStephenspon – “Was it fear of socialism, or nationalism?”

    It was both – but above all it was fear that Ed Miliband was too soft to prevent a party with just 59 seats stomping all over his party with circa 250 seats.

    Think about the coalition of the Conservatives with the LibDems. People always looked at the LibDems as the Junior Partner. Input yes, dictats no. Conservative voters felt Cameron fought their corner.

    Some Lab voters didn’t feel that way about Ed Miliband.

    Paradoxically if he’d really been the Ruthless B***ard that stabbed his brother in the back, he’s have had no problems. Unfortunately he was the soft guy that 16-year old girls felt they had to defend, and adult voters felt he’d be swallowed whole by La Sturgeon.

    Do this thought experiment: imagine Dan Jarvis (Mr Operation Honey Badger) being the Labour leader – could you imagine the SNP dictating to him in a coalition? If no, then the that line of attack (being controlled by the SNP wouldn’t have worked with him as leader).

    So at the end of the day, the problem was Miliband.

1 9 10 11 12 13 14