1. We have added a Gift Upgrades feature that allows you to gift an account upgrade to another member, just in time for the holiday season. You can see the gift option when going to the Account Upgrades screen, or on any user profile screen.
    Dismiss Notice

French presidential election 2017

Discussion in 'Off-Topic' started by AdrienIer, Nov 20, 2016.

  1. Akka

    Akka Moody old mage.

    Joined:
    Nov 14, 2001
    Messages:
    13,367
    Location:
    Facing my computer.
    It's rather ridiculous and annoying to talk at length about betting when the thread is about politics.
     
  2. Leoreth

    Leoreth Prince of Blood Moderator

    Joined:
    Aug 23, 2009
    Messages:
    34,486
    Gender:
    Male
    Location:
    House of Hades
    Yeah that's what I meant, betting markets for some reason have entered into the realm of politics discussion because some people think they can serve the same purpose as opinion polls.
     
  3. Kraznaya

    Kraznaya Princeps

    Joined:
    Sep 28, 2005
    Messages:
    6,822
    Location:
    Land of the Successor
    Betting markets are a lagging indicator for the polls, not a leading one. So I agree that treating them with any degree of predictive power is silly, however, Americans tend to enjoy horse race mechanics and treat politics as a bit of a spectator sport. Some might find it a bit distasteful but I think the discussion at least underscores where we think the race is heading.
     
  4. Bootstoots

    Bootstoots Deity Retired Moderator

    Joined:
    Mar 2, 2003
    Messages:
    9,437
    Location:
    Mid-Illinois
    There was a short time (c. 2014-16) where people thought that betting markets might provide more accurate feedback than poll aggregation and (especially) punditry because having an actual financial stake in the outcome has been shown to make people more objective in their analysis in some psychological studies. But now this has been largely discredited; turns out betting markets just reflect what bettors think is going on.

    They do however provide a decent feel for what the group of bettors thinks the probability of a given outcome is. Which may not be particularly useful for actually predicting an election, but it sheds some light on how humans think about probability and the political future. Having a candidate who is down 20 in a long and consistent series of polls still be given a ~15% chance, shortly before election day, indicates that some people think that any right-wing populist upset is possible, regardless of what the polls say. This is of course a fundamental misreading of the situation, and after Le Pen loses by something like the expected margin, they'll probably stop behaving that way to quite the same extent.

    A lot of what people got out of the Brexit and Trump phenomena, both of which were only likely but nowhere near certain to break the other way according to the polls, is that polls and pre-election data in general are useless. I strongly suspect this is not correct - what actually happened is that pundits treated narrow margins for the establishment to be equivalent to a sure victory, in part because they couldn't imagine the alternative. Now we've been seeing people overcorrect and think that right-wing populism is sweeping the world, but at this point they're probably substantially overestimating it. If Le Pen were getting odds in the low single digits instead, I'd think the narrative of right-wing populism sweeping the world was ringing hollow, at least among the sorts of people who bet on elections. I can see numerically that it's not, though, and a big chunk of people think that any right-wing populist upset is possible. Betting markets help to quantify where people believe political trends are going.
     
  5. innonimatu

    innonimatu Deity

    Joined:
    Dec 4, 2006
    Messages:
    12,667
    I don't think there will be withdrawals in the assembly elections this time. France no longer has only two big parties, more candidates will get to the second round and they are not going to quit.
     
  6. Grisu

    Grisu Draghetto Retired Moderator

    Joined:
    Dec 20, 2001
    Messages:
    10,505
    Gender:
    Male
    Location:
    Switzerland
    Ah yes, thanks. I forgot about that one. IIRC during the last elections several regions pulled their PS candidates in the second round. So it's expected that the FN gets a similar share of the votes than in the first round presidential elections?
     
  7. AdrienIer

    AdrienIer Deity

    Joined:
    Mar 4, 2010
    Messages:
    3,379
    Location:
    Paris
    No one knows. The participation is usually lower (around 55% in 2012) and nobody knows who will come out and vote.
    In the last elections the FN voters were not as lazy as the others (hence their 25% national score in the first round of the regional election) but will they keep being active this time ?

    Edit :
    I disagree. The front republicain is alive and well for the most part, as this election has proven. Except in districts where there are personal grudges between the candidates, most will withdraw if the other option is the FN.

    The tradition on the left has always been that if two candidates from the left are in the second round the one that finished behind withdraws in the second round. Even in stupid situations like when there are no other candidates in the second round. For example in my district 5 years ago the communist candidate withdrew after the first round, leaving the socialist as the only candidate.
    I doubt they'll be as kind this year, but there's still some room between what happened last time and having no withdrawals at all.
     
    Last edited: May 5, 2017
  8. Kyriakos

    Kyriakos Alien spiral maker

    Joined:
    Oct 15, 2003
    Messages:
    58,962
    Location:
    Thessalonike, The Byzantine Empire
    Pollsters also have a financial stake in the outcome; they are just pre-paid, though :D
     
  9. Kyriakos

    Kyriakos Alien spiral maker

    Joined:
    Oct 15, 2003
    Messages:
    58,962
    Location:
    Thessalonike, The Byzantine Empire


    He does like older women, as known :jesus:
     
  10. Leoreth

    Leoreth Prince of Blood Moderator

    Joined:
    Aug 23, 2009
    Messages:
    34,486
    Gender:
    Male
    Location:
    House of Hades
    Not very logical, if they are pre paid why do they have financial stake in the outcome.
     
  11. Kyriakos

    Kyriakos Alien spiral maker

    Joined:
    Oct 15, 2003
    Messages:
    58,962
    Location:
    Thessalonike, The Byzantine Empire
    Not very logical, if a poll is posted it can influence actual voting and thus the outcome.
     
  12. metalhead

    metalhead Angry Bartender

    Joined:
    Apr 15, 2002
    Messages:
    8,031
    Pollsters' financial stake overwhelmingly lies in being accurate, at least for well-known public pollsters. If you can't produce results that match the outcome, you won't find people willing to pay for your polling numbers.

    This is why pollsters tend to herd, especially towards the end of an election cycle. If you end up where everyone else is, you don't run the risk of being singled out for having bad polls. So yeah, their financial stake can absolutely affect their outcomes, but in the same way anyone else's would - they produce results to A. stay in business, and B. be accurate. Usually doing B. will facilitate A., but A. will win out every time they are faced with a choice between trusting their numbers and protecting their business.

    However I feel like it is very poorly understood that neither Trump nor Brexit was a particularly bad polling miss. Leave had pulled just about even right prior to the referendum and was leading in many of the polls, and Trump lost by 2 points nationally where the polls had him losing by 3. The state polling was off on Trump by significantly more in PA, MI, and WI, which is why how he won seemed so surprising, but these supposed "polling misses" weren't really misses at all.

    I think this is because people, particularly people who are of a mind to try to analyze polling data, were genuinely shocked by the outcomes and therefore the polling must not have prepared them for it, but that's not true. Leave was consistently winning in multiple polls through the entire time that FT tracked polls. The fact that it won should not have been a huge surprise.
     
  13. Kyriakos

    Kyriakos Alien spiral maker

    Joined:
    Oct 15, 2003
    Messages:
    58,962
    Location:
    Thessalonike, The Byzantine Empire
    ^Well, to a crucial degree my own comment included sniping ;)
    That said, it isn't unlikely that even main/brand pollsters have less than savoury ties to some establishment interests.
     
  14. Bootstoots

    Bootstoots Deity Retired Moderator

    Joined:
    Mar 2, 2003
    Messages:
    9,437
    Location:
    Mid-Illinois
    Yeah, the Brexit "miss" was entirely on the pundits: the polls consistently showed it within the margin of error. I think they must have run into some sort of inconceivability bias where they couldn't imagine voters not thinking like they do, and magnified the tiny Remain advantage into something that seemed like a sure thing in their heads.

    The national polls were right on Trump, but with large regional variability so that they underpredicted him in the Midwest while actually overpredicting him in electorally irrelevant states like California, Texas, and Massachusetts. As we know now, the reason 538 did much better than other models is that they had neighboring states and states in the same region correlate strongly with each other, so that a loss in Michigan would also mean Clinton was in trouble in PA and WI, while being crushed in OH.

    Even then, though, the polls' worst miss in a state that actually flipped was in Wisconsin where the RCP average was Clinton up by 6.5 percentage points while the real outcome was Trump by 0.8, for a 7.3% error. Bumping that up to 8%, the French equivalent would be Macron's expected 60-40 win turning into 56-44 instead, which wouldn't really affect anything. French pollsters do seem to herd quite strongly, but even that plus abstention by some left-wingers won't flip this for Le Pen.
     
  15. Akka

    Akka Moody old mage.

    Joined:
    Nov 14, 2001
    Messages:
    13,367
    Location:
    Facing my computer.
    I don't really think Le Pen can win (though you have to consider that abstention and blank votes will play a MAJOR role), but if she manage to pull over 40 % (and even more if she reaches 45 %), it might still be pretty seismic even as a loss.
    The problem, as usual, is that everyone talk about taking into account the anger and changes that are expressed by "extreme" votes, but nobody actually do jack about it, and it's business as usual - until it's too late. So we can hope it's a wake-up call, but I'm pretty pessimistic about it.
     
  16. Lillefix

    Lillefix I'm serious. You can.

    Joined:
    Dec 1, 2003
    Messages:
    5,688
    Le Pen should be in more debates

     
  17. Akka

    Akka Moody old mage.

    Joined:
    Nov 14, 2001
    Messages:
    13,367
    Location:
    Facing my computer.
    She certainly didn't give a positive image of herself in the previous one.
     
  18. Takhisis

    Takhisis Jinping, wer fragt uns?

    Joined:
    Jul 11, 2005
    Messages:
    50,754
    Location:
    up yours!
    And then the Jeunesse du Turin attacked.
    And as you post further down, less than 40% in the electorate turned up for the latter.

    All the others were within the margin of error.
    Whenever I hear the word ‘patron’ I automatically think of Jules Maigret. Blame Simenon for that.
    Yes, except that many people bet on whomever they're voting for as a show of loyalty.
     
  19. Bootstoots

    Bootstoots Deity Retired Moderator

    Joined:
    Mar 2, 2003
    Messages:
    9,437
    Location:
    Mid-Illinois
    Looking at this page, it appears that not a single poll out of 23 taken since the first round has predicted Macron to get anything outside the range 59-63%, and the Ifop-Fiducial poll was the first to predict 63%, only doing so today. Each of these polls should have margins of error (1 standard deviation) between 1.5 and 3%, so there should be several polls outside that 4-point range.

    Now that's what I call herding! I'd love to know how all the pollsters are adjusting their numbers to make sure they always get very close to 61-39.
     
  20. AdrienIer

    AdrienIer Deity

    Joined:
    Mar 4, 2010
    Messages:
    3,379
    Location:
    Paris
    Small reminder : french polls are almost never done with a random sample. That eliminates some of the result randomness.
     

Share This Page