538 Model Megathread
       |           

Welcome, Guest. Please login or register.
Did you miss your activation email?
March 29, 2024, 08:16:42 AM
News: Election Simulator 2.0 Released. Senate/Gubernatorial maps, proportional electoral votes, and more - Read more

  Talk Elections
  Election Archive
  Election Archive
  2016 U.S. Presidential Election
  538 Model Megathread
« previous next »
Pages: 1 ... 29 30 31 32 33 [34] 35 36 37 38 39 ... 49
Author Topic: 538 Model Megathread  (Read 82821 times)
Crumpets
Thinking Crumpets Crumpet
Atlas Icon
*****
Posts: 17,586
United States


Political Matrix
E: -4.06, S: -6.52

Show only this user's posts in this thread
« Reply #825 on: October 24, 2016, 12:35:13 AM »

Somebody did this earlier on, so I thought I'd make an update.

Using polls only:

Trump's 99.9th percentile:


Trump's 99th percentile:


Trump's 90th percentile:


Trump's 87th percentile (narrowest potential Trump win):


Clinton's 14th percentile (narrowest potential Clinton win):


Both candidates' 50th percentile:


Clinton's 87th percentile (this map is equally likely as a Trump win):


Clinton's 90th percentile:


Clinton's 99th percentile:


Clinton's 99.9th percentile:


Logged
Dr. Arch
Arch
Atlas Icon
*****
Posts: 12,452
Puerto Rico


Show only this user's posts in this thread
« Reply #826 on: October 25, 2016, 03:55:30 PM »

I just realized that another model that Nate could use would include early-voting turnout/expectations alongside the factors included in Polls-Plus, something like Polls-Premium?

I would certainly like to see how elements such as the extremely positive turnout in states like FL, TX, and AZ would pull things one way or another. What do you guys think?
Logged
Mr. Morden
Atlas Legend
*****
Posts: 44,073
United States


Show only this user's posts in this thread
« Reply #827 on: October 25, 2016, 04:05:38 PM »

I just realized that another model that Nate could use would include early-voting turnout/expectations alongside the factors included in Polls-Plus, something like Polls-Premium?

I would certainly like to see how elements such as the extremely positive turnout in states like FL, TX, and AZ would pull things one way or another. What do you guys think?

Everything in the model (both polls and the "poll plus" variables) is based on the predictive value of the parameters as established by previous presidential elections.  Has early voting really been around long enough to tell you how much predictive power those things have on the election outcome?  You have to be able to say "X% of the time where this variable says Y, the leading candidate wins."  But if early voting has only been around for one or two presidential elections, then you don't have a large enough sample to be able to say anything.
Logged
Dr. Arch
Arch
Atlas Icon
*****
Posts: 12,452
Puerto Rico


Show only this user's posts in this thread
« Reply #828 on: October 25, 2016, 04:08:03 PM »

I just realized that another model that Nate could use would include early-voting turnout/expectations alongside the factors included in Polls-Plus, something like Polls-Premium?

I would certainly like to see how elements such as the extremely positive turnout in states like FL, TX, and AZ would pull things one way or another. What do you guys think?

Everything in the model (both polls and the "poll plus" variables) is based on the predictive value of the parameters as established by previous presidential elections.  Has early voting really been around long enough to tell you how much predictive power those things have on the election outcome?  You have to be able to say "X% of the time where this variable says Y, the leading candidate wins."  But if early voting has only been around for one or two presidential elections, then you don't have a large enough sample to be able to say anything.


A fair point. Perhaps several cycles from now, it will be a possibility.
Logged
Erich Maria Remarque
LittleBigPlanet
YaBB God
*****
Posts: 3,646
Sweden


Show only this user's posts in this thread
« Reply #829 on: October 26, 2016, 08:57:08 AM »

http://fivethirtyeight.com/features/election-update-where-are-the-undecided-voters

Logged
Gass3268
Moderators
Atlas Star
*****
Posts: 27,478
United States


Show only this user's posts in this thread
« Reply #830 on: October 26, 2016, 09:09:40 AM »

Quote
You must be logged in to read this quote.
Logged
Dr. Arch
Arch
Atlas Icon
*****
Posts: 12,452
Puerto Rico


Show only this user's posts in this thread
« Reply #831 on: October 26, 2016, 09:19:45 AM »


In short, no.
Logged
Beefalow and the Consumer
Beef
Junior Chimp
*****
Posts: 9,123
United States


Political Matrix
E: -2.77, S: -8.78

Show only this user's posts in this thread
« Reply #832 on: October 26, 2016, 09:34:38 AM »


Take away Nate's "unskewing" and the race is almost back to pre-pussygate levels (2-way):

Logged
Gass3268
Moderators
Atlas Star
*****
Posts: 27,478
United States


Show only this user's posts in this thread
« Reply #833 on: October 26, 2016, 09:41:19 AM »


Take away Nate's "unskewing" and the race is almost back to pre-pussygate levels (2-way):



RCP is biased on what polls they put in their averages. Republican firms like Remington are fine, but not Democratic ones like PPP or Democracy Corps.
Logged
muon2
Moderators
Atlas Icon
*****
Posts: 16,788


Show only this user's posts in this thread
« Reply #834 on: October 26, 2016, 10:26:58 AM »
« Edited: October 26, 2016, 10:35:41 AM by muon2 »

If you use HuffPo 3-way and limit the polls to nonpartisan pollsters and likely voters the current numbers from their trend lines are Clinton 45.1%, Trump 40.7%, Johnson 5.1%. For the 2-way it gives 45.8% to 42.4% (moderate smoothing)
Logged
The Other Castro
Castro2020
Atlas Icon
*****
Posts: 11,230
United States


Show only this user's posts in this thread
« Reply #835 on: October 26, 2016, 10:35:14 AM »

McMullin's chance of winning Utah is now over 30% in the Nowcast.
Logged
ursulahx
Jr. Member
***
Posts: 527
United Kingdom


Show only this user's posts in this thread
« Reply #836 on: October 26, 2016, 11:53:22 AM »

Has the race tightened? Yeah, slightly. Will it make a difference? Not much.
Logged
Maxwell
mah519
Atlas Star
*****
Posts: 28,459
Germany


Political Matrix
E: -6.45, S: -6.96

Show only this user's posts in this thread
« Reply #837 on: October 26, 2016, 12:40:30 PM »

Clinton's numbers aren't going down. Trump still hasn't crossed 45% in the two-way.
Logged
riceowl
riceowl315
YaBB God
*****
Posts: 4,350


Show only this user's posts in this thread
« Reply #838 on: October 26, 2016, 03:54:51 PM »

In polls-only, Clinton's odds of winning NV = Trump's odds of winning UT Wink
Logged
dspNY
Sr. Member
****
Posts: 2,802
United States


Show only this user's posts in this thread
« Reply #839 on: October 26, 2016, 04:47:50 PM »

That AP/GFK poll caused Clinton to spike 2 points in the Nowcast, 1.7 points in polls-only and 1.4 points in polls plus Smiley

They recorded that poll as Clinton 51, Trump 37, Johnson 6
Logged
Antonio the Sixth
Antonio V
Atlas Institution
*****
Posts: 57,965
United States


Political Matrix
E: -7.87, S: -3.83

P P
Show only this user's posts in this thread
« Reply #840 on: October 26, 2016, 05:29:04 PM »

That AP/GFK poll caused Clinton to spike 2 points in the Nowcast, 1.7 points in polls-only and 1.4 points in polls plus Smiley

They recorded that poll as Clinton 51, Trump 37, Johnson 6

That's the sign of a model that's way too sensitive.
Logged
Devout Centrist
Atlas Icon
*****
Posts: 10,102
United States


Political Matrix
E: -99.99, S: -99.99

P P
Show only this user's posts in this thread
« Reply #841 on: October 26, 2016, 05:30:09 PM »

That AP/GFK poll caused Clinton to spike 2 points in the Nowcast, 1.7 points in polls-only and 1.4 points in polls plus Smiley

They recorded that poll as Clinton 51, Trump 37, Johnson 6

That's the sign of a model that's way too sensitive.
No, it just reversed the 'tightening' people here were having mini heart attacks about this morning.
Logged
Antonio the Sixth
Antonio V
Atlas Institution
*****
Posts: 57,965
United States


Political Matrix
E: -7.87, S: -3.83

P P
Show only this user's posts in this thread
« Reply #842 on: October 26, 2016, 05:33:12 PM »

That AP/GFK poll caused Clinton to spike 2 points in the Nowcast, 1.7 points in polls-only and 1.4 points in polls plus Smiley

They recorded that poll as Clinton 51, Trump 37, Johnson 6

That's the sign of a model that's way too sensitive.
No, it just reversed the 'tightening' people here were having mini heart attacks about this morning.

That's part of my point. It overreacted to the polls showing a tightening and now it's overreacting to a single poll showing a widening.
Logged
Devout Centrist
Atlas Icon
*****
Posts: 10,102
United States


Political Matrix
E: -99.99, S: -99.99

P P
Show only this user's posts in this thread
« Reply #843 on: October 26, 2016, 05:34:43 PM »

That AP/GFK poll caused Clinton to spike 2 points in the Nowcast, 1.7 points in polls-only and 1.4 points in polls plus Smiley

They recorded that poll as Clinton 51, Trump 37, Johnson 6

That's the sign of a model that's way too sensitive.
No, it just reversed the 'tightening' people here were having mini heart attacks about this morning.

That's part of my point. It overreacted to the polls showing a tightening and now it's overreacting to a single poll showing a widening.
Well Nate has a huge hard on for Selzer, so I think that's what started this whole mess.
Logged
dspNY
Sr. Member
****
Posts: 2,802
United States


Show only this user's posts in this thread
« Reply #844 on: October 26, 2016, 05:37:50 PM »

Arizona is now ahead of Iowa on all three models and is almost blue on polls-plus (49.8%)
Logged
Erich Maria Remarque
LittleBigPlanet
YaBB God
*****
Posts: 3,646
Sweden


Show only this user's posts in this thread
« Reply #845 on: October 26, 2016, 05:45:25 PM »

That AP/GFK poll caused Clinton to spike 2 points in the Nowcast, 1.7 points in polls-only and 1.4 points in polls plus Smiley

They recorded that poll as Clinton 51, Trump 37, Johnson 6

That's the sign of a model that's way too sensitive.
1-2% is not so sensitive in my world. Though it would look better, if they didn't show the decimals.
Logged
Ozymandias
Jr. Member
***
Posts: 470


Show only this user's posts in this thread
« Reply #846 on: October 26, 2016, 06:32:18 PM »
« Edited: October 26, 2016, 06:38:44 PM by Ozymandias »

Nate Silver ‏@NateSilver538  1m1 minute ago
Now for the world's most boring tweet: Here's how 538 model uses or doesn't use various Electoral College projections from online pollsters:

https://twitter.com/NateSilver538/status/791421640609497089

Nate Silver ‏@NateSilver538  3m3 minutes ago
Basically, some of these are Electoral College projections are polls, and some are models. We use the polls but not the models.
Logged
afleitch
Moderators
Atlas Star
*****
Posts: 29,837


Show only this user's posts in this thread
« Reply #847 on: October 27, 2016, 02:42:43 PM »

538 adds the Quinnipiac polls, adjusts them up in Clinton's favour and Trump's chances bounce up. He's up from 13.7 to 16.9 on the now cast over the course of the day. I find that unusual.
Logged
15 Down, 35 To Go
ExtremeRepublican
Moderators
Atlas Icon
*****
Posts: 11,669


Political Matrix
E: 7.35, S: 5.57


Show only this user's posts in this thread
« Reply #848 on: October 27, 2016, 02:47:28 PM »

538 adds the Quinnipiac polls, adjusts them up in Clinton's favour and Trump's chances bounce up. He's up from 13.7 to 16.9 on the now cast over the course of the day. I find that unusual.

I think their logic was that these polls confirm that IA and NC are within range if Trump surges nationally (the only way he will win) before November 8.
Logged
Erich Maria Remarque
LittleBigPlanet
YaBB God
*****
Posts: 3,646
Sweden


Show only this user's posts in this thread
« Reply #849 on: October 27, 2016, 02:51:55 PM »
« Edited: October 27, 2016, 02:54:10 PM by Happy Sad Trumpista »

538 adds the Quinnipiac polls, adjusts them up in Clinton's favour and Trump's chances bounce up. He's up from 13.7 to 16.9 on the now cast over the course of the day. I find that unusual.

I think their logic was that these polls confirm that IA and NC are within range if Trump surges nationally (the only way he will win) before November 8.

Sort of. The model think Hillary is up 6% nationally (and for example C+2 in IA etc). So relatively to this, Trump got a good polls. But those are still awfull for him.

And Pew was included as well. They adjusted it to +5, which is a "good" result for Trump.
Logged
Pages: 1 ... 29 30 31 32 33 [34] 35 36 37 38 39 ... 49  
« previous next »
Jump to:  


Login with username, password and session length

Terms of Service - DMCA Agent and Policy - Privacy Policy and Cookies

Powered by SMF 1.1.21 | SMF © 2015, Simple Machines

Page created in 0.219 seconds with 13 queries.