Thursday, June 8, 2017

The case for pessimism...and the case for mild optimism

One of the great dramatic rituals of British democracy will play out tonight when, at the exact moment Big Ben strikes 10, the main broadcasters simultaneously reveal the results of their jointly-funded exit poll.  (In days gone by, the BBC and ITN each used to have their own exclusive exit poll, but now they put their combined eggs in one basket.)  If past form is anything to go by, the results of the poll will probably not be leaked in advance - although we can be sure that the Neil Lovatt brigade will be interpreting every little movement in the betting markets from the afternoon onwards as a sign that somebody has found out something.

There are two key points that need to be borne in mind about the exit poll -

1) There is a significant chance that it will throw up a major shock.  That's happened in all of the last three general elections.  The 2005 exit poll forecast Tony Blair's majority to be "only" 66 seats, which was significantly lower than most people expected.  In 2010, it was predicted that the Cleggasm had melted away and left the Liberal Democrats with fewer seats than they started with - something that hardly anyone had entertained as a serious possibility.  And in 2015, expectations were completely turned on their heads as the Conservatives were predicted to have achieved a net gain, Labour were predicted to have suffered a net loss, and the SNP were forecast to have taken all but one seat in Scotland.

2) If a shock is predicted, it's highly likely to be reflected in the actual result.  In 2005, Labour ended up with exactly the 66 majority that was suggested.  And I'm sure we all remember Paddy Ashdown simply refusing to accept the exit poll's pessimistic estimates for the Lib Dems in both 2010 and 2015, and how he was later forced to eat his words.

So this is my worry.  We're all a little bit too cosy in our assumptions about the minimum number of seats we think the SNP are practically guaranteed to win.  Based on previous shocks, we should really be braced for almost any number to flash up on the screen - it could be below 40, in theory it could even be below 30.  And if the prediction is much worse than expected, we'll almost immediately have to reconcile ourselves with the likelihood that we're not looking at a 1992-style dud poll, but rather at a reasonably accurate prediction.

It's not too hard to construct a case for why the SNP might underperform expectations.  The polls we saw last night basically put the SNP in the same position - or perhaps a marginally worse position - than they were in at the time of the local elections.  None of us need any reminders that the SNP's lead over the Tories in those elections was only 7%, rather than the 10-15% that the polls would have led us to anticipate.  If you factor in the impact of independent candidates standing in areas where the Tories are strong, it's arguable that the SNP's de facto lead on the day was even lower than 7%.  Now, we've tended to assume that the SNP only suffered in those elections because of particularly extreme differential turnout - Tory supporters were whipped up into a frenzy over the issue of an independence referendum, while the SNP fought a rather bland campaign that didn't motivate their own core support.  I still think that's an extremely plausible theory...but what if it's wrong?  What if, plain and simply, the polls have been systemically and significantly overstating the SNP's true lead over the Tories?  In that case, all bets would be off for tonight.

That's the pessimistic side of the coin, but here's the optimistic side.  A couple of threads ago, Calum identified some of the SNP-to-Labour switchers among his own peer group - young people who support independence, but who wrongly believe that a vote for Labour is the most effective way of getting the Tories out.  I have a sneaking suspicion that if those people lived in Moray or in Perth & North Perthshire, they would be well aware by now, either from leaflets or through word of mouth, of the folly of that way of thinking in the context of the local battle.  It's just possible that the polarisation of this campaign along Tory v anti-Tory lines is actually working in the SNP's favour in Tory-SNP battleground seats, with natural Corbyn supporters swinging heavily behind the SNP.  If that's happening, the swing to Labour must be disproportionately taking place in ex-Labour heartlands - but even if that's true, the numbers might still be insufficient for Labour to take back very many seats from the SNP.  They're starting from such a long way back almost everywhere.

So that's the scenario in which it's feasible that the SNP might just about get the strong mid-40s result that everyone seems to be expecting - but if it does happen, it looks like it could be a bit of a tightrope-walk.

*  *  *

As long-term readers know, I have a great regard for the predictive powers of Stephen Bush (in spite of the brief slanging-match I had with him after the Brexit referendum).  So I've been waiting with bated breath to hear his verdict on the suggestions that English Labour are doing much worse on the ground than the opinion polls suggest.  Intriguingly, he departs from the conventional wisdom in saying that the Labour surge is real, and that he's found evidence of it from speaking to local organisers.  However, he also thinks the extra votes are very inefficiently distributed and that Labour still face a net loss of seats even if their vote share increases markedly.

In a perverse way, I take some heart from that assessment.  If there's to be even the remotest chance of a hung parliament, the number one prerequisite is that the change in public opinion detected by the polls at least has to be real.

54 comments:

  1. What's you gov seat predictor saying?

    ReplyDelete
  2. C'mon James don't go all concern troll on us...nothing less than 55 seats tonight and you know it

    ReplyDelete
    Replies
    1. Tories in scotland were predicting they would win 19 seats a few weeks ago.

      We'll see if that passes the 'smell test' after tonight.

      Delete
    2. Someone (maybe you) made this claim a few threads back. I asked for a link and didn't get a reply. Can you provide one?

      Delete
    3. This comment has been removed by the author.

      Delete
    4. I don't think there's going to be a meltdown in the SNP seat's either I think the Nats tend to do better when written off. I think we can hold the same number, and maybe increase as the unionist tactical voting seems to be dividing and conquering which is very ironic considering the role of sectarianism in maintaining the union. I would laugh if the SNP increase its numbers and only lost to Patrick Harvey.

      Delete
    5. Seems we​ got both our forecast wrong oh well still over 50% of the
      Seats in Scotland, I guess if we were a unilateral independence movement we could off avoided this.again!

      Delete
  3. Final election polls give Jeremy Corbyn the lead over Theresa May for the first time
    http://www.wired.co.uk/article/election-polls-labour-conservative-winner
    There is always hope...................

    ReplyDelete
  4. The more polls, the more uncertainty. We have to assume there will be losses, though the only reason you can identify is the hope amongst NE fishermen that English Tories have restoration of the fishing limits the sine qua non of negotiations. And undoubtedly there are those who think May will improve attainment in Scottish primary schools.

    ReplyDelete
    Replies
    1. undoubtedly there are those who think May will improve attainment in Scottish primary schools.

      Haaaahahahahahahahahahah!!

      Delete
  5. In perth labours canvassing was no existant, i didnt get a single leaflet.

    Torys have been very active, were out at 630 this morning leafleting.

    Im off to helpout with getting our vote out now.

    ReplyDelete
    Replies
    1. I believe Perth is one of the constituencies where Labour agreed not to campaign in the hopes of giving the Tories the seat.

      Delete
  6. What's the predictions UK wide as well?

    ReplyDelete
  7. Something is happening in UK politics and I think the pollsters are only half picking it up. Despite relentless attacks from the media and his own party Jeremy Corbyn is attracting a lot of support. We don't know how many or how likely they are to vote or whether they are in the right constituencies to make a difference. You would think polling ought to be better now than it ever was, but the population seems to be capable of voting in a way the experts don't predict. Think Brexit, Trump. They were disastrous decisions I think, but on the whole I like the idea of voters not being easy to manipulate.

    ReplyDelete
    Replies
    1. Unfortunately, I think too many are easy to manipulate. We have debunking websites with rebuttals and intelligent arguments, but we have seen that simplistic slogans and propaganda still have some effect in Scotland.

      Delete
  8. The Herald online says that the BMG poll was 'conducted over the last two days', although your article suggests on out of date field work. If this is correct it is rather more favourable to SNP - suggesting they will take 55 seats. With a big Tory majority also predicted by them, that is surely a good recipe for Scottish independence. It would be good to have your take on this

    ReplyDelete
    Replies
    1. As far as I'm aware, it was The Herald's GB-wide BMG poll that was conducted over the last couple of days.

      Delete
  9. There will be some very close races if there is much unionist tactical voting. If the SNP achieve around 40-41% as these polls seem to be suggesting then on sheer averages their share will fall into the 30s in many seats. Where there is one other strong party in the seat, it must at that point become a race for every vote.

    In my most optimistic moments I think 50 seats. But having looked at current majorities etc in every seat, I would in no way be astonished if the SNP ended up with 35 (I hope to be wrong, but that's an assessment with my dispassionate head on - it just doesn't look impossible if an ill wind's blowing).

    ReplyDelete
    Replies
    1. Well one of my tory mates, yes I know, has just voted Lib Dem in a seat that is between SNP and tory. So that will provide a laugh for you.

      Hope there is a few other tories not on the ball.

      Delete
    2. All the same, Johnny, looking at the range of bookies on Oddschecker, they still have the SNP odds-on favourites (or greater odds-on where the contest is tighter) in several of the seats touted as the most vulnerable. Would so many bookies be wrong in all of these? Which leads one to anticipate that other seats are safe. If the bookies are right, we can expect 48/49 seats.

      Delete
    3. Bookies' prices reflect where the money is going from punters, rather than a polling prediction.

      Only time it's their 'guess' is when they set the odds at the start of the event.

      If the odds haven't moved, it might be because the markets are not taking much money, as opposed to meaning they are winning each seat.

      Delete
    4. The stockmarkets, with all the data and number-crunching powers at their disposal, called both Brexit and Trump wrongly. Markets closed on those days on the complacent assumption these events weren't going to happen, a few hours later as the actual results unfolded, they went into a nosedive. Almost makes me think that no-one knows anything about anything...

      Delete
    5. What Johnny said.

      Imagine you had 2 candidates that were basically neck and neck, we'll call them Blue and Red. Now imagine that you had 2 bookmakers. The first bookie gets a lot of bets on Blue, while the second bookie gets a lot of bets on Red.

      Do you think the two bookies would show the same odds?

      The answer is, of course, no. The Blue bookie would want to attract some red bets, while the Red bookie would want to attract some blues. So they'd tweak their odds accordingly.

      Remember, bookies make money on the spread. Ideally they want a balance of bets so they can't lose. They don't like to gamble, they leave that to the punters.

      Delete
  10. We're going to turn Scotland into blue cheese.

    ReplyDelete
    Replies
    1. "A 2005 survey carried out by the British Cheese Board reported that Stilton cheese seemed to cause unusual dreams when eaten before sleep, with 75% of men and 85% of women experiencing "odd and vivid" dreams after eating a 20-gram serving of the cheese half an hour prior to sleeping."

      And I've been thinking that some things round here don't pass the smell test. Still, let's see what happens with the crumblies.

      Delete
    2. Thats just between your toes Aldo

      Delete
  11. My thoughts on constituency results for the SNP.

    50+ smashing
    45-50 acceptable given its Westminster vote
    40-45 concerning, need to review campaigning strategies and policy priorities
    under 40 disastrous, need to radically change approach to confronting the unionist threat.

    ReplyDelete
  12. Been out getting the vote out and polling station duties in South Perthshire the tories i have spoken to think they are going to clean up Perth North and South many think SNP seat forecast mid to high 20s. Thats what they think hope they are hopelessly wrong but they are definitely going about like a dog with two cocks.

    ReplyDelete
    Replies
    1. Well, of course, if they feel confident of taking that seat (and it's certainly one that's up for grabs) they themselves might be falling into the fallacy of thinking 'it must be like this in every seat, near enough. Everyone thinks like me'.

      Or they're kidding on they think that.

      Who knows?

      Delete
  13. The tories aren't going into another election with May as leader after her incredible clusterfuck of a campaign.

    ReplyDelete
  14. YouGov haven't updated their model today, which I was interested to see would show a swing to the Tories. Hedging their bets perhaps? If it's a Tory majority they can point to how their final poll was right, and if it's a hung parliament they can point to their fantastic model.

    ReplyDelete
  15. The final YouGov model update is coming tonight after the polls close.

    ReplyDelete
    Replies
    1. Along with all the other edit polls....

      Delete
  16. Had a look round a lot of political forums and very little SNP poster's posting. Are things that gloomy?

    ReplyDelete
    Replies
    1. With a bit of luck they're all busy getting the vote out.

      Delete
    2. Yes, we are. Drookit but determined.

      Delete
  17. The last exit poll in 2015 had the SNP on 58 seats and they won 56. What ever tonight's poll says I would take about 5 off the total due to tactical voting. I don't think exit polls can pick up tactical voting. Im going for the SNP to win 43 seats

    ReplyDelete
    Replies
    1. It should be able to pick up tactical voting - it'll certainly aim to do that. The reason the SNP's seats were overestimated last time is simply that their vote was overestimated (I think the exit poll had them in the low 50s).

      Delete
    2. The 2015 exit poll definitely had the SNP on 58 seats.

      Delete
    3. No, I meant the exit poll had them in the low 50s in the popular vote! I think it was 52% or something like that. That's why they were predicted to win 58 seats.

      Delete
    4. Sorry - got you now.

      Delete
  18. Turnout in my area rural Aberdeenshire is around 60-65%

    ReplyDelete
  19. Not long till the exit poll. Gulp. I'm not at all confident about my prediction, but hey, might as well mention it one final time just in case it's correct:
    Snp 50 ConservaSpivs 8 FibDooms 1 Slab zilch.

    ReplyDelete
  20. In line with 2015.

    So assuming the same people voted as in 2015....snp should be fine if they gotv.

    ReplyDelete
    Replies
    1. Should add, obvs snp has lost votes but I dont think there has been that much of a swing for snp to lose.

      @billpalmer

      Delete
    2. Which seat are you in, Chalks? Presumably Gordon or Banff & Buchan?

      Delete
    3. Banff and Buchan

      Heard an area of Gordon has 70% snp exit poll...obvs carried out by snp polling agents outside..

      Delete
  21. For the SNP to be reduced to 34 MPs is a disaster imo. Major questions for Sturgeon if this is accurate imo.

    ReplyDelete
  22. Chris Cairns‏ @cairnstoon 33 minutes ago

    Pollster on @SkyNews just admitted the range of SNP total in exit poll is from 21 to 50. In other words #fuckinguseless

    ReplyDelete