Five thirty eight dot com

Discussion in 'Politics, Religion, Social Issues' started by TheAppleFairy, Nov 8, 2016.

  1. TheAppleFairy macrumors 68020

    TheAppleFairy

    Joined:
    Mar 28, 2013
    Location:
    The Clinton Archipelago unfortunately
    #1
    Suddenly their site is looking different. Don't mean to gloat (or maybe I do)
    --- Post Merged, Nov 8, 2016 ---
    This was their prediction 4 hours ago
     

    Attached Files:

  2. AustinIllini macrumors demi-god

    AustinIllini

    Joined:
    Oct 20, 2011
    Location:
    Austin, TX
    #2
    FiveThirtyEight adjusts their predictions based on called states.

    If any blog/website won this election, it's fivethirtyeight. Most sites didn't give Trump a chance.
     
  3. Zombie Acorn macrumors 65816

    Zombie Acorn

    Joined:
    Feb 2, 2009
    Location:
    Toronto, Ontario
    #3
    I think we learned that polls are useless (again)
     
  4. TheAppleFairy thread starter macrumors 68020

    TheAppleFairy

    Joined:
    Mar 28, 2013
    Location:
    The Clinton Archipelago unfortunately
    #4

    I understand that I just love seeing the sudden change. How many times over the past several months have people thrown up the screenshots from the site showing that Trump has no chance of winning?
     
  5. Populism macrumors regular

    Populism

    Joined:
    Jun 11, 2014
    #5
    Search my name on this forum, and you'll find at least four or five posts at least where I praised FiveThirtyEight as the gold standard of polling.

    Wow. Joke is on me. Guess I'm getting my comeuppance.

    Nate - I drank your koolaid, you doucher.
     
  6. AustinIllini macrumors demi-god

    AustinIllini

    Joined:
    Oct 20, 2011
    Location:
    Austin, TX
    #6
    They are the gold standard of polling. Polling was just wrong this time around.
     
  7. aaronvan Suspended

    aaronvan

    Joined:
    Dec 21, 2011
    Location:
    República Cascadia
    #7
    Nate Silver: enjoy your unemployment, you asshat.
     
  8. TheAppleFairy thread starter macrumors 68020

    TheAppleFairy

    Joined:
    Mar 28, 2013
    Location:
    The Clinton Archipelago unfortunately
    #8
    I want to search for all the post that said "we will never have a republican president again" or "the Republican Party is done".

    Well this election isn't over but there is definitely some new momentum even if trump doesn't win.
     
  9. APlotdevice macrumors 68040

    APlotdevice

    Joined:
    Sep 3, 2011
    #9
    We learned that Americans as a whole are incredibly gullible.
     
  10. A.Goldberg macrumors 68020

    A.Goldberg

    Joined:
    Jan 31, 2015
    Location:
    Boston
    #10
    Pretty much no one got the sophisticated polls right, with maybe the exception of the LA-Times poll which was considered to be really out there. 538's model is dynamic by design, but clearly it is flawed. 538 is derived from baseball statistics, but let's face it, a political election is so much more complex and dynamic than baseball stats.

    Polling, like a research survey, is highly unreliable. There's probably no one size fits all forumula to predict an outcome.
     
  11. unlinked macrumors 6502a

    Joined:
    Jul 12, 2010
    Location:
    Ireland
    #11
    Wasn't the LA Times an unusual methodology where they chose a group to track and then polled the same people through the whole election. Sounds interesting but obviously sensitive to getting it right when you pick people.
     
  12. A.Goldberg macrumors 68020

    A.Goldberg

    Joined:
    Jan 31, 2015
    Location:
    Boston
    #12
    Correct. They were a consistent outlier amongst all other major polls. I'd suggest entirely possible aka more than likely it was simply a coincidence. I believe they had a 3000 subject sample size they followed throughout the election season rather than switching the sample group every poll.

    Polling samples generally speaking are not very big, 3000 isn't particuarly large. When you compare 3000 versus 100m voters, there's a lot of room for error. So again, it may just be coincidence.

    Maybe there is something to be said about sticking with the same subjects in an election like this. Trump support wasn't exactly a socially acceptable opinion considering all his negative press. Psychologically speaking, virgin pollers might be more likely to lie in polling even though it really means nothing- particuarly if they're polled talking to a real person. If someone is being polled regularly by the same organization, they might be more willing to be honest about their socially frowned upon candidate.
     
  13. Snoopy4 macrumors 6502a

    Snoopy4

    Joined:
    Dec 29, 2014
    #13
    What I thought was interesting was the overall methodology of following the same people. It was interesting to see how they moved back and forth. I don't know if that makes a good predictor of outcome, but probably tells much about how people feel about ecpxternalities that affect the election.
     
  14. Teon macrumors regular

    Teon

    Joined:
    Feb 14, 2016
    #14
    predictions have credibility. Otherwise, there would be no election - all have been based on predictions
     
  15. DearthnVader macrumors 6502

    DearthnVader

    Joined:
    Dec 17, 2015
    Location:
    Red Springs, NC
    #15
    538 was a bunch of hooey, but the underlying data they got from the polls they were using was flawed.

    I pointed out a few times that Democrats were being oversampled in quite a few polls, skewing the results. I can understand how it is appealing to poll 60% Democrats in a state like PA, where the registered voters are 60% Democrat, but that doesn't gage if those 100% of those people are going to vote, and vote Democrat.

    I'd say we're going to have a postmortem of the polls, hopefully they will learn to adjust their skews the proper way.
     
  16. v0lume4 macrumors 68000

    v0lume4

    Joined:
    Jul 28, 2012
    #16
    FiveThirtyEight has the nicest-looking site around, but when the data you run your calculations off of is flawed to begin with, your results are wrong.

    This election has been a smear campaign against Trump from the beginning, with polls showing him down in the general election since as far back as I remember. Ridiculous.

    Power to the people. Viva Democracy!
     
  17. Herdfan macrumors 6502

    Joined:
    Apr 11, 2011
    #17
    Actually I don't think they were oversampled, but instead overweighted. Actually they may have been undersampled in that many Dems crossed over in states like MI, PA and WI. Most polls didn't catch that.

    Looks like many were close nationally, but at the state level where it mattered they blew it. I think they also underestimated the willingness/eagerness of Trump supporters to get out and vote just like they did with Obama in 2012.
     

Share This Page