Friday, June 24, 2016

Memo to Theresa Villiers : You're on a loser with this one

I was just watching Channel 4 News, and midway through it, Theresa Villiers (Northern Ireland Secretary and Leave campaigner) said something truly extraordinary. First she said that she agreed with Ruth Davidson that now is not the right time for an independence referendum, and then she stated that Scotland had voted to remain in the United Kingdom at a time when it already knew that a UK referendum on EU membership was on the cards.

Well, that's very curious, because I suspect the people of Scotland might just remember that Ruth Davidson herself, in a high-profile TV debate during the independence referendum, informed voters that her own party were unlikely to win the 2015 general election (let alone win an overall majority), and that we could therefore vote No safe in the knowledge than an EU referendum would NOT happen. I'm not sure how the position can be much clearer - the narrow No vote was won specifically on the premise that it would safeguard our EU membership, rather than imperil it within less than two years.

I'm afraid Ms Villiers will have to learn to understand that agreeing with the leader of the Scottish Tory party is not a part-time occupation.

It's official : Scotland is to be dragged out of the European Union against its will

Unfortunately I had to close the previous thread because of an abusive anonymous commenter, but feel free to continue the discussion below.  A few miscellaneous points on this extraordinary morning...

* Against all the expectations, online polls have emerged as more accurate than telephone polls in this campaign.  (Although Populus was a huge outlier in that regard, and it's been almost as bad a night for Populus as it was for David Cameron.)

* The 'wisdom of crowds' theory in relation to polling has been busted.  That was the idea that you get a more accurate forecast if you ask people to predict the result, rather than give their own voting intention.  In fact, voting intention polls have proved to be much more accurate - all of the prediction questions went heavily in Remain's favour.

* Never again will anyone be remotely impressed by the argument that movements on the currency and betting markets are a better guide to a likely election result than public opinion polls.

* * *

I had an unexpectedly eventful night, because I was interviewed live on Canadian TV just after midnight. It's only the third time I've been interviewed on the mainstream media, and this was completely different from the two previous occasions because it was done from the comfort of my living room via an iPad that I frantically had to ask to borrow at very short notice. I was able to give a little context about how differently the referendum was being viewed in Scotland compared to the rest of the UK, and also how for many people Westminster is the problem rather than Brussels.

We may be heading for a narrow Leave win - and for Indyref 2

There may yet be a joker in the pack tonight, as there was in the 1997 Welsh devolution referendum when the BBC briefly forecast No as the winners, only for Carmarthenshire to completely buck the national trend and carry Yes over the line with an overwhelming vote.  But at the moment, Leave appear to be the more likely winners.  We heard strong indications from Brian Taylor earlier (at a point when it still looked like Remain would coast to victory) that in the event of a vote to withdraw, Nicola Sturgeon would at least very quickly demand that Westminster gives Scotland the power to hold an early referendum if it so chooses.

I don't know about the rest of you, but I hope there's no backtracking on that plan - I think that's absolutely what's got to happen.  Voters were told in 2014 that if they voted No, they were voting to remain EU citizens.  They duly voted No.  Today, Scotland voted by a more than 'decisive' margin (to use the approved BBC word for anything north of 54%) to remain EU citizens.  We must now be allowed to remain EU citizens.  If that doesn't happen, it would be a democratic outrage - it's as simple as that.

UPDATE : I should point out that Remain are still the favourites on Betfair.  I can't say I remotely understand that given the numbers we're all looking at, but perhaps I'm missing something and the scenario I've just outlined isn't likely to happen after all.

UPDATE II :  I'll tell you another thing about what's happening tonight, and I know some of you don't want to hear this - it absolutely obliterates the myth that Yes would need to have a strong lead in the polls before it would have any chance of winning an independence referendum.  There isn't always a reversion to the status quo in constitutional referendums, and that must now be abundantly clear.

Thursday, June 23, 2016

YouGov on-the-day poll predicts narrow Remain victory

Should the United Kingdom remain a member of the European Union or leave the European Union?

Remain 52% (+1)
Leave 48% (-1)

This is very reminiscent of the YouGov on-the-day poll for the indyref, which showed the tiniest of last-minute swings in favour of the status quo.  The difference is, though, that Remain are on 52% compared to 54% for No two years ago, which leaves open the potential for a Leave victory if YouGov are just a little out.  As it happens, they were a little out two years ago, but it was Yes that they overestimated by 1%.  We'll see.

UPDATE : Ipsos-Mori's on-the-day phone poll looks more convincing for Remain -

Remain 54% (+2)
Leave 46% (-2)

So it does look like there's been a genuine late swing towards the status quo, albeit not a massive one.  Leave can only really win now if there's been some kind of systemic problem with the public polls - although that's scarcely unheard of.

UPDATE II : To give a tiny glimmer of hope to Leave, it looks like the YouGov figures have flattered Remain slightly due to rounding - it's actually Remain 51.6%, Leave 48.4%.

Good evening Edinburgh, this is Budapest calling. Here is how the Hungarian jury would have voted tonight.

As we approach the end of polling, it's interesting to note that Stephen Bush of the New Statesman is still saying that he thinks we might be looking at a Leave vote. That should be taken seriously, because he was one of the first to confidently predict that Jeremy Corbyn would win the Labour leadership, and he was also closer to the truth than most about the general election. However, his assessment about tonight is based on an assumption of relatively low turnout. The mood music on social media actually suggests a very high turnout, although there's weirdly contradictory information out there if you look hard enough - different people in London saying turnout is either abysmally low or heading for a record-breaking high.

Turnout in London is particularly important, as is turnout in Scotland and Northern Ireland, because those are likely to be Remain's three best areas (unless you count Gibraltar). Between them they make up about one-quarter of the UK's population.

Anyway, as we wait for that mystery to be solved, I thought I'd treat you (and this won't take long) to an abortive interview I did a couple of weeks ago with my Hungarian friend Anita, who has lived in Scotland for most of the last seven years. As you might expect, she's not best pleased about the fact that she has no influence over a decision that will arguably affect her more than the British and Commonwealth citizens who do have a vote. I was going to ask her fourteen questions and really give her a chance to sound off, but unfortunately circumstances intervened and it came to an abrupt end just after I asked Question 2.

Me : You were about fourteen years old when Hungary joined the European Union in 2004. Do you have any memories of that period?

Anita : Yeah, I have loads of them. The first thing we all thought was that we would have the euro, you know, join the eurozone. That didn't happen! But I knew that there were all these new possibilities, that we could move, that we could travel, and you wouldn't need a passport. You could just go anywhere and see the world. I'd never been abroad before, so it was something that I really wanted to do. And everyone wanted to do that.

Me : Did you notice anything different about Hungary after it joined the EU?

Anita : Prices went up, I noticed that. People started to leave...

And...well, yeah, that was it. Not quite Frost/Nixon, but clearly I have to walk before I can run.

Remain take narrow lead in final Ipsos-Mori telephone poll

If you're ever trying to organise a surprise party, it might be an idea not to mention it to Ben Page of Ipsos-Mori.  Having told the world the other day that two of his firm's competitors were about to publish polls showing Remain in the lead, he's spent the last couple of hours undermining the embargo on his own poll by dropping broad hints on Twitter that it was going to show a swing to Remain.  In all honesty, though, given that Leave have been dropping like a stone on the betting markets, it's quite striking that the Remain lead is as low as it is - it's lower than the ComRes phone poll reported last night (in spite of the fact that the fieldwork is much more up to date), and it's in 'statistical tie' territory.  Depending on the reports they're getting about turnout on the ground, the Leave campaign may feel this still puts them in with a genuine shout.

Should the United Kingdom remain a member of the European Union or leave the European Union?

Remain 49% (+6)
Leave 46% (-3)

UPDATE : The Populus poll is much better for Remain -

Remain 55%
Leave 45%

That's a real oddity, because it's an online poll, and the fieldwork is only marginally more up to date than the YouGov and Opinium online polls that had a roughly 50/50 split. There have been some suggestions that Populus (who haven't published an EU poll for months) have been doing the private polling for the Remain campaign, so if those polls have been quietly doing the rounds in elite circles and have been markedly better for Remain all along, that could explain the odd sense of certainty in some quarters about today's outcome. But it's worth remembering that private polling isn't inherently more accurate than private polling. It'll be interesting to see whether Populus have been doing anything unusual, and perhaps reassigning Don't Knows to Remain on the basis of answers to supplementary questions in the way that ComRes did last night.

UPDATE : We now have a bit more detail from the Ipsos-Mori poll (although not the full datasets yet). The most intriguing point is that when a slightly different turnout filter is used, the Remain lead drops to 2% - a 51% to 49% split with Don't Knows excluded.

It's human nature to react to polls in the order in which we see them, but it should be stressed that the Ipsos-Mori poll is just as up-to-date as the Populus poll - both were conducted yesterday and the day before. And of course Ipsos-Mori is a telephone poll, so in the ultimate role-reversal it's actually the telephone poll that's keeping Leave's hopes alive.

Wednesday, June 22, 2016

Remain open up 6% lead in final ComRes phone poll

Should the United Kingdom remain a member of the European Union or leave the European Union?

ComRes (phone) :

Remain 48% (+2)
Leave 42% (-3)

YouGov (online) :

Remain 51% (+2)
Leave 49% (-2)

The ComRes result is, I'm sure, going to be portrayed in some quarters as a hammerblow for Leave, but actually, a quick perusal of past ComRes results reveals this to be the second-best poll of the series for Leave - the Remain lead has only slipped below 7% on one previous occasion, and that was in the most recent poll.  The other big saving grace for the Leave campaign is that the ComRes fieldwork took place over several days, and actually started as long ago as Friday - so these numbers aren't any more up to date than any of the other polls we've seen today or yesterday.  They're an important piece of evidence, but they don't supersede what has gone before.  Basically this returns us to the long-running conundrum of this campaign - do you trust phone polls more than online polls, or is the truth somewhere in between?

* If online polls are calling this correctly, we're heading for a cliffhanger.

* If phone polls are right, the odds favour Remain - although there have been fewer phone polls to go on, and the three most recent ones have given conflicting estimates of the size of the Remain lead (arguably the ORB poll managed to contradict itself).

* If the true position is at a rough midway point between online and phone polls (which seems to be John Curtice's view), you'd very slightly favour Remain, but you certainly wouldn't be betting your house on it - particularly given that it's so hard to make predictions about how differential turnout will play out in an unprecedented one-off referendum.

My namesake James in the comments section below has pointed out (and I'm taking his word for it, because the YouGov website has crashed!) that the percentage changes in the YouGov poll are misleading, because there's been a methodological change - a 'squeeze' has been added to take account of undecided leaners.  On a like-for-like comparison with the previous poll and excluding the squeeze, the figures are apparently Remain 45% (+3), Leave 45% (+1).  A dead heat is within YouGov's recent normal range, so as was the case with Opinium and TNS (and I would argue Survey Monkey as well), there's absolutely no hard evidence of a very recent swing to Remain in this poll.

Another commenter raised a point that we've heard a lot about in recent days - that by some mysterious process, the City somehow already "know" that Remain will win, and will receive between 52% and 54% of the vote.  This was my reply, and I think it bears repeating -

"Short of buying into the conspiracy theories and accepting that elections in Britain are rigged, that just sounds like gibberish to me (either that or professional arrogance). Even if you were looking at extensive private polling evidence showing that the Remain vote was around 52-54% with Don't Knows excluded, that does not constitute hard evidence that Remain is definitely going to win, because it's close enough for Leave to sneak it if the turnout-modelling is wrong. (And given the uniqueness of the contest, why would anyone have 100% confidence that the turnout modelling is well-founded?)"

The issue of turnout modelling is particularly relevant to the ComRes poll, because the ComRes adjustment for turnout has tended to be more favourable for Remain than what other pollsters have been doing. The basic figures among the whole sample tonight are Remain 46.8%, Leave 42.1%.

ComRes have also rather boldly reassigned some Don't Knows to Remain on the basis of how they answered a supplementary question about the economy. That could be a dangerous thing to do, given that the economy and immigration have been vying for importance throughout this campaign. Time will tell whether it was a stroke of genius or a foolish 'reimagining' of the raw results of a poll. Without that highly unusual adjustment, the headline Remain lead would have been a touch lower.

If it's Brexit, what then for Scotland?

Just a quick note to let you know that I have a new article at the International Business Times about the potential impact of a Leave vote tomorrow on Scottish politics : the powers that would be automatically transferred from Brussels to Holyrood, the constitutional crisis that might ensue if Westminster tried to breach the Sewel convention by clawing some of those powers back, and of course the very real prospect of an early second independence referendum.  You can read the article HERE.

Leave cling on to the lead in final TNS poll - just

Should the United Kingdom remain a member of the European Union or leave the European Union?

Remain 41% (+1)
Leave 43% (-4)

From what I can gather, TNS have chosen this unlikely moment to make a big methodological change, which entirely explains the apparent movement towards Remain.  Without it, Leave would have retained exactly the same 7% lead they had in the last poll from the same firm.  That's pretty extraordinary, given that the previous poll was conducted when Leave were enjoying a bit of a purple patch across the whole of the polling industry.  On the face of it, that bodes well for a Leave victory, because it implies there's unlikely to have been much of a swingback to Remain - but the snag is that the fieldwork was conducted over a relatively extended period.  It finished today, but it started way back last Thursday.  That leaves open the possibility that there has been a very recent swing to Remain that this poll was unable to detect.

UPDATE : Just to emphasise the point about there being no underlying swing to Remain in this poll, I've now had a look at the datasets, and it appears that among the whole sample (when undecided 'leaners' are not assigned to either side) the Leave lead has actually increased from 2% to 4%.

Leave take slender lead in final Opinium poll

I'm sure we'll have more polls tonight (Referendum Eve is very much like Christmas Eve, as veterans of the indyref will recall), but we've already had two so far today - and the news is mixed.  Leave will be delighted to take a 1% lead with Opinium, who in recent weeks have become one of the most Remain-friendly online firms after introducing methodological changes that essentially made their weighted sample more like a typical telephone sample.  But Survey Monkey show a small swing in the opposite direction, which leaves open the obvious possibility that nothing much has changed, and that any apparent shifts in either poll are just caused by normal sampling variation.  Both polls show a statistical tie - ie. even if the methodology is bang on accurate, it's not possible to say which side is in the lead due to the standard margin of error.  (I gather Opinium may have used a larger sample than usual, but even enormous samples have a margin of error.)

Should the United Kingdom remain a member of the European Union or leave the European Union?

Opinium :

Remain 44% (n/c)
Leave 45% (+1)

Survey Monkey :

Remain: 50% (+2)
Leave: 47% (-2)

I've just received the Survey Monkey press release by email, in which they describe the small reported shift to Remain as "decisive".  I must say that's a bizarre summary, and not just because Opinium have directly contradicted their findings.  Even with a sample size of 4000, it's absurdly bold to suggest that a 2% swing in a stand-alone poll is necessarily significant, or indeed that a 3% gap means that you've definitely got the correct side in the lead.  I suspect that Survey Monkey, just like BMG a few days ago, may be overstating their sense of certainty in the hope of gaining bragging rights for years to come if they just happen to be right.