Jump to content

The Dog

Burnt Ends
  • Posts

    10,277
  • Joined

  • Last visited

  • Days Won

    56

Posts posted by The Dog

  1. 5 minutes ago, BradInATX said:

     

    Yeah, if you want some "it just scares you to death" action, here you go: https://luckboxmagazine.com/trends/this-pollster-says-the-polls-are-wrong-trump-will-win/

    Interview with the Trafalgar guy. He thinks 2016 is going to happen again. I think he's wrong and I think like you said his methodologies are making false assumptions that 2020 is going to do the same exact thing as 2016, but it's enough to keep a bit of fear in me. 

    He's not wrong about social desirability and some of the labels branded to Trump supporters.  It's especially true for college educated whites where they know they are in the minority and don't particularly want to be outcasts because of their political preference(s). 

    I also agree that some of these polls are waaay too long. That's one thing you'll notice about the USC poll is it's short and sweet and it's a tracking poll with a large sample size. 

    Now where I think he's wrong is that social desirability works both ways. The press has written about the possibility of a "shy Biden voter" in traditionally red states where Trump supporters are much more open and vocal. I also think the idea that the "shy Trump voter" effect is stronger this year is very wrong. It may be there but it isn't like 2016.

    • Hook 'Em 1
  2. 27 minutes ago, Pato del Muerto said:

    Now I’m really questioning the accuracy of morning consult. 

    That's why you look at the averages.

    For example, Trafalgar just dropped a poll that showed Trump +2 in Michigan and Fox shows Biden +11. Both are clear outliers as Trafalgar is likely under-sampling Biden support and Fox is under-sampling Trump support (Fox has Trump at only 40% in their poll and I don't buy that at all). 

    So look at the average and it's Biden +8 with Biden averaging nearly 51% in all polling. RCP average is just slightly lower but in line with 538. 

    That's likely where things stand. Remember that 4 years ago the pollsters really only missed Trump's support due to the large number of late deciders and the media not paying attention to the fact that the races in the midwest and elsewhere were within the margin of error. 

    • Hook 'Em 1
  3. 1 minute ago, Ghost of LL said:

    Different election; different candidates.

    Really, I don't find that instructive at all.  The thing that stands out about 2016 was how volatile it was.  And to some degree, that's to be expected when you don't have an incumbent.

    What marks 2020 is how steady the race has been.  Trump has never had a very good chance of winning.  And since September 1, his chances have just gotten worse and worse.

    Yeah I wasn't intending to say "OMG 2016 ALL OVER AGAIN" - people need to look at 2016 for what it was and recognize 2020 is different. 

    Additionally we have a LOT more votes that have already been cast (as you and I have noted) and so any late surprises just can't have the same impact. That and the small number of undecideds. 

  4. 16 minutes ago, Storm the Field said:

    CNN polls not too shabby for Joe, though they have unusually large LV/RV splits.

    PA 53-43-2 LV, 50-45-2 RV

    FL 50-46-3 LV, 48-46-3 RV

    Possible that their LV screen is overweighting DEMs that have already voted early in those states. 

    Regardless, Biden strong position in PA, and not enough undecideds/third-party to really matter. 

    I read that as meaning Democratic turnout exceeds their original likely voter model and that Republican respondents aren't as likely to vote. Usually Republicans have the edge in likely voter models, but now with this much time to vote early it's much easier to model likely voter turnout, because they have already voted.

  5. 3 minutes ago, Ghost of LL said:

     

    And respectfully to the Nates and G. Elliott, I trust John Cornyn's prognostication abilities more than I trust theirs.  He has been counting votes for a long time.  And his career is riding on this.  If he thinks Donald Trump is losing, I'm going to trust him.

    It's because Cornyn has access to GOP internal polling that Nate and G. Elliott do not. They may hear about internal polls second hand through media sources but he gets to see the raw data.

    • Hook 'Em 1
  6. 20 minutes ago, bolverk said:

    Here's a map of the turnout as of yesterday according to the SOS website. The Hill Country, Wilco, North DFW suburbs, are coming on strong.

    Statewide, 5.3 million have already voted which puts us at 31.3 percent turnout so far.

    Early-Voting.jpg

     

     

     

    Got a link to this? 

  7. This tool would be better if you could adjust the probabilities of an individual state. It's either definite Biden/Trump win or whatever they choose based on Silver's probabilities. It would be more fun to adjust the probabilities of an individual state (like Florida - which Silver indicates is a key to determining the winner) and see the impact on the overall election. It would allow me to adjust Florida to Trump being "slightly favored" and seeing the impact rather than Trump DEFINITELY winning and the impact.

×
×
  • Create New...