Jeremy's Den: The Scottish Play

Going Home 1979, Hearthammer 1991, Siol Ghoraidh 1989

A few days before I published my latest Westminster update, a fellow Scottish tweep suggested I do a simulation on the effects of low turnout in Scotland at the Christmas Shopping Spree Election. Which was actually something I had in mind but right then I was doing another kind of simulation, about the effect of the Unionist vote coalescing around the Conservatives, which you can see in my previous article. Then the differential turnout idea stuck, because it is a good one, and also because it looks like a better way to motivate SNP voters, the couch-slouching-on-Irn-Bru-drip ones at least, to get the fuck out to vote. Which they did in 2015 and did not in 2017, for all the reasons I have already listed. And the tone of the current campaign proves beyond doubt that the SNP top brass reached the same conclusion and will concentrate on the one BIG issue (guess you have already figured out which one) and leave rambling about devolved matters during a GE campaign to Scottish Labour, who still haven't read the whole of the series of Scotland Acts. Then another of my followers suggested I made my articles more 'digestible', which he meant in a good way like in 'shorter and fucking focused, lad' and I have to say I tend to agree as I sometimes have a hard time finding out where the fuck I was actually going when I reread past articles. And longish sentences without punctuation don't help either. So here we go, a shorter one, focused just on what could possibly go wrong in Scotland on 12 December. And that fucking lengthy and undigestible intro is also a fine excuse to squeeze in another Runrig video before I move on to the heart and liver of the matter. Especially as the extra one is The Story from their very last studio album, which I rate as one of the most brilliant songs the MacDonald brothers ever wrote. Sure you don't mind.... Or do you? Really?

Where are we now?

When I started writing this article three days ago, I started by whining about having only one recent full Scottish poll, which was (and still is) one month old already. Which was a major change from 2017 when we already had seven Scottish post-dissolution polls at the same point in the campaign, and also from 2015 when we had five. Then Big Ben struck thirteen and Skye slid beneath the waves of The Mighty Atlantic, when out of the blue came that now infamous Panelbase poll that has probably already shattered the whole SNP campaign strategy. I had also prepared projections based on Scottish subsamples of GB-wide polls, which I usually don't do for all the obvious reasons already stated by me and others. So don't take this as more than a fun exercise on what polls say, or don't. So here is what my model projects from YouGov's and Panelbase's Scottish polls, YouGov's and other pollsters' subsamples and the aggregate of all subsamples over the last three weeks. Note that all the projections have one thing in common: they predict that Stephen Gethins will lose North East Fife to the LibDems. I don't think this will ever happen but I kept is as delivered in the name of mathematical purity, even if it goes against my hunches and political likelihood. Then I guess PM Jo's catabysmal trainwreck on Question Time will also nosedive the LibDem vote generally as they lose yet some more of the goodwill votes they had squirrelled over the last few months.

I singled out YouGov here partly because they have the most frequent pollster miles and also because they have become James Kelly's pollster of choice recently for some good reasons and others not so good. One of the latter is obviously that they are the most SNP-friendly of all. You can certainly argue, like James does, that their greater backlog and better methodology make them mor reliable. Or you could argue that some unknown skews in their demographics deliver some SNP bias here as they deliver some Tory bias in their England results. There is some massive irony here in YouGov totally missing what the Panelbase poll found, while the aggregate of other pollsters' subsamples matched it. Just wondering if James will revise his profiles of Scottish constituencies for the National, though they were largely non-committal enough to not need revising whatever the polls say. So for now it might be wiser to follow the more pessimistic views, because there is a fucking hell of a difference between gaining seven seats and gaining sixteen, and not just because one is more than double the other. A lot of difference in the way the English MSM can spin it, the Conservatives can spin it and even the SNP themselves can spin it. If the Tories hold more than ten seats as Panelbase and some others predict, they can easily argue that there is not that much appetite for Independence after all (more later to disprove this), and that blue rosettes remain the best shields against the assaults of the demonic separatists. And the SNP gaining seats only or mostly from Labour would just signal yet another realignment within the left-wing electorate, which has already happened twice, or so we were told. So for starters, after recovering from the shockwave of the Panelbase poll, I updated my earlier simulation about the Unionist vote coalescing around the Conservatives.

What we have here on updated vote shares is not that much different from what I got the first time. There are just fewer steps as the Conservatives start from a much higher baseline that earlier with the YouGov poll. And again the SNP would still retain a majority of seats in a 40%-40% situation. Which probably would not impress the Unionists that much anyway. But it is actually not the main point here as this simulation deals only with reshuffling the votes within the Unionist camp. So it says nothing about the consequences of a lower turnout or whatever patterns of differential turnout. More on this after the Runrig interlude.

Loch Lomond © Traditional, 1841ish or about
Hearts Of Olden Glory © Calum MacDonald, Roderick MacDonald, 1987
The last songs of the last show of The Last Dance tour

What could possibly go wrong?

Now let's go to my first differential turnout simulation. I will keep this one simple with only one variable: the number of SNP votes. Originally I intended to take the baseline voting intentions from the last YouGov poll, only slightly tweaked. The poll was conducted before the Brexit Party stood down in Tory-held constituencies, so I intended to change the baseline data from 22% Tory and 6% Brexit to 25% Tory and 3% Brexit, which looks like a good idea at the time and a more realistic assessment of changing voting intentions. Then came the Panelbase poll, which also changes the baseline and the seat projection to an even less favourable starting point for the SNP. Which better fits the whole purpose of the exercise, which is to evaluate the worst case scenarios for the SNP in a realistic contexte. Hypothetically I factored in here the same overall turnout as in 2015 (2,910k voters), to be on the optimistic side of things. The SNP on a predicted 40% would then bag a baseline of 1,156k voters and the Conservatives an already ominous 802k.

The next step is easy: incrementally decrease the number of SNP voters while keeping all the other parties on the same number of votes, until we reach an overall turnout in the vicinity of 2017's, when 2,650k voters showed up at their polling place, and then one more step beyond to test a possible carcrash worst case scenario on just 2.6 million voters. Here the SNP ends up well under 1 million voters (2017 result was 968k votes) and has its vote share reduced to somewhere in the low thirties. As you might expect, this has a huge impact on the predicted distribution of seats.

The Conservatives' first gains; when we reach a 2;754k turnout, would be East Lothian from Labour, then Lanark and Hamilton East, Perth and North Perthshire from the SNP, who would have already lost North East Fife to the LibDems., At the next step, the Conservatives would also snatch Argyll and Bute, Central Ayrshire and Edinburgh South West from the SNP. Ironically the last two steps would see Labour recovering a wee smitch, 'unlosing' Coatbridge, Chryston and Bellshill first and Midlothian next. This somewhat simplified version shows which SNP seats could be at risk and which expected gains are the weakest, which I expect Surgeon Central and the local branches already know. There is a weakness here as the whole simulation relies on just one variable and the real situation would be more complex, and potentially much more risky. So get braced for something more sophisticated and much more disturbing after Runrig's tribute to the Scots who lost everything in the Clearances.

And what could possibly go even wronger?

The mechanics here become more complex as I factor in four variables: the vote shares for the SNP, Labour, the LibDems and the Conservatives. To do this I gathered data on the likelihood to vote by voting intentions as published in YouGov's October poll, as Panelbase did not publish similar data in their November poll. Then this is all a theoretical construct, so YouGov's data are as good as any, even if I find them quite optimistic as they predict an overall turnout approaching 80%. But this reflects the motivation of voters who have already chosen one of the four major parties, and you can expect minor parties' voters and undecideds to be less motivated and likely to vote. And as usual in this kind of exercise, the differences between the various hypotheses matter more than the raw numbers per se. Then I recalculated the SNP's votes step by step as if the baseline reflected a high turnout including even the moderately motivated, those rating 5/10 on the likelihood to vote scale, and shrunk it until it included only the core highest motivated, those rating 10/10 on the likelihood to vote scale. And recalculated both the Conservative and Labour votes with the exact opposite hypotheses. Finally the LibDem vote becomes kind of the adjustment variable here as I decreased it when the Labour vote rose, on the somewhat risqué assumption that a number of moderately left-leaning LibDem voters could switch to Labour to help oust Johnson.

When you push the worst case scenario to the worstest, you get something that would make Jackass Carlaw spunk his Y-fronts faster than it would take Jo Swinson to answer a straight Yes-Or-No-To-Ministerial-Cars question. Then I'm not saying any of the trainwreck scenarios (the three bars on the right) could actually happen, but take a closer look at the third bar from the left. What we have here implies, by sheer coincidence, a combination of events that I would not totally rule out just yet. A dose a voter apathy resulting in the SNP and LibDems doing poorly and holding just their 2017 vote shares. A dose of Unionist coalescence around the Conservatives including a 3% swing from Labour to the Conservatives and a barely visible Brexit Party vote. And this is one case in which I did not push the variables too hard, unlike in the next and more unbelievable scenarios. It's just two steps away from the baseline delivered by the Panelbase poll and the factors that lead to such a result are far from extravagant. So don't say it absolutely can't happen. Or be ready for one fuck of a serious electoral hangover on Friday the Thirteenth.

In this scenarios the road to 19 Conservative seats matches was I said earlier. I will spare you the gory details of the most extreme one as it certainly belongs in fantasy land. But something in the mid-range definitely could happen if we're not paying attention enough to people who might still think it's in the bag already. Which is the most powerful key to unfavourable differential turnout this year when the polls show the SNP doing less well than anyone anticipated just weeks ago. That's the reason why, even without looking at differential turnout scenarios, the SNP must invest in a massive GOTV drive and aim for a much larger number of votes than in 2017. Aiming for a return of all 2015 voters is probably over-ambitious, but there is no reason why the sky shouldn't be the limit. Or is there?

Now what about IndyRef2020?

As they say every shitestorm has a silver lining and the Panelbase poll is no exception. They polled voting intentions for the next Independence referendum and found 45.5% Yes, 47.3% No and 7.2% who don't know or wouldn't vote. Which translates into a neat 49% Yes to 51% No once you've deported the dontknowdontcarers to Penrith, so statistically as much of a tie as earlier polls that went 50-50 or 51-49. This last result is totally in line with the trend of Independence polls since September 2014, that shows increased support over the last two years after a noticeable slump in 2017. But we're not yet past the hurdle though we probably would be by now if the SNP had not totally fucked up the issue during the 2017 GE campaign. Now this campaign is offering another golden opportunity to make the case and it should not be missed. Obviously the SNP understand this as they have again propelled Independence at the top of their list of talking points. Honestly I don't think we will have IndyRef2 on Nicola Sturgeon's timetable, that is next year, but there is a real chance we'll have it soon after the next Holyrood election if we play all our cards cleverly in the meantime. And naw I don't mean the way Tory James would, we're better than that, aren't we?

The weighted average of the three most recent Independence polls is just as clear as the overall trend. A Yes-No tie that could go either way when you factor in the margin of error. Now we wouldn't want IndyRef2 delivering 47% Yes, or would we? Sure that would be better than 2014 but would turn any further attempt into a once-in-seventeen-generations thing. So some steps have to be taken first. Not coincidentally ICM included a question about the Independence referendum in their last voting intentions poll, because they think it has become a major issue in the present campaign. 40% of respondents GB-wide support having IndyRef2 and 42% oppose it, with 18% undecided. Which is kind of interesting as it shows England is not radically against the idea but borderline irrelevant. Most importantly 49% of Scottish respondents support it and 46% oppose it with only 5% undecided. This once again kills the Unionist talking points about 'no appetite', in case they actually care and are not stuck in a Groundhog Day of denial. And in case Jeremy Corbyn still finds it convenient to procrasturbate on the issue, he should know that 56% of Labour voters support IndyRef2 with only 28% opposed and 17% undecided. But of course all this matters only if the triple-lock mandate is secured. Step One: an undeniable success at the Christmas Shopping Spree Election, an in case you haven't noticed, this could be harder than expected. Step Two: Both Votes SNP and a SNP majority of seats at the next Holyrood election, so we won't have to worry about the Little Green Men's next backstabbing. Leafleting, canvassing and GOTV. Plus Nicola on TV. And this will work.

Panelbase also explored other aspects of the Independence and Brexit debates. 37% of respondents think Independence would be a greater threat than Brexit for the Scottish economy and 45% think Independence would offer a greater opportunity. Conversely 39% think that Brexit would pose a greater threat and 24% that it would offer a greater opportunity. Which leaves a high proportion of people who don't have an opinion either way but also offers an oven-ready talking point to the SNP for this campaign and others to come. Simply make the case on how much Brexit would hurt Scotland especially in the already most deprive areas and of course hammer it home in Tory-leaning areas in the North East in case they haven't realized yet how much they were betrayed by the party they support. Simultaneously stress how many new doors Independence would open especially with the opportunity to join/rejoin an existing customs union and/or single market, and keep a certain level of constructive ambiguity about whether it would be the EU or EFTA. Just in case you might want to convince the pro-Indy pro-SNP Leave voters that there is still life outside the EU, just outside, or wouldn't we want to make that case too? From where I stand, we should. But that's a story for another day, after we win.

And since I indulged in a bit of Runrigstalgia, did I ever tell you that I always felt there is definitely something of John Bonham in Iain Bayne? Aye, never mind. Next general update on Thursday E-14. Spoiler: won't look good for Labour.

Don't be afraid of anything, don't be afraid of anyone

Where are the frontiers? How do I get there?

There are no maps, there are no roads

Just you and me on these streets of gold

(Paul Kantner and Marty Balin, America, from KBC Band, 1986)

Jeremy's Den

25/11/2019

The Scottish Play - Act V of MMXIX

No comments:

Post a Comment

In Every New Poll A Headache