Post

Conversation

Nate Silver
@NateSilver538
I'll put this chart in front of the paywall today just for transparency's sake. When you're near 50/50, a 1-point shift in the polls = about 8% of win probability shift in November. So easy to end up on either side of the line on any given day.
Image
524.2K
Views
David Watson 🥑
Post your reply

Nate Silver
@NateSilver538
Another illustration of this is how our *popular vote* forecast has basically been within Harris +1 to Harris +2.5 the entire time. Small changes in state polls can make a big difference in Electoral College probability.
Image
160K
Jake Snake
@JakeSnake4321
Lmao bro you’re going to have egg on your face again like you did in 2016…you’re feeding GARBAGE TRASH polls into your model….garbage in garbage out
1K
Jason Morton
@jamorton0419
Wouldn't time, i.e. closer to election day, also have a greater impact on these shifts also? For example, a 1-point shift today would be less dramatic than a 1-point shift two weeks from now? So we could see wilder swings in the model the closer we get to election day?
916
AX
@A_XENOP
A lot of these recent pollsters have a very spotty record the last cycles. How do you assess these polls and their methodology, and how you incorporate it into your model and output? Because right now a lot of the polling seems to be a bit schizophrenic.
133
David Goldiner
@davidgoldiner
Tx for your transparent explanation ... but it actually reveals how crazy flawed your model is because it responds to statistically insignificant shifts with HUGE movements in the chance of winning (in either direction)
383
American Desi Anonymous
@AmrikanDesiAnon
You seem not to be dealing with the elephant in the room. Your whole electoral college win probability based on popular vote margin is being contradicted by the relationship between national polls and swing state polls this cycle.
211
Rob Polinsky
@poman927
Interesting that with all the polls yesterday, your model increased Harris from a 49% to 51% chance while the 538 model decreased Harris from a 64% chance to 60% chance. Are the models starting to converge as the event approaches, kind of like weather with the GFS & Euro models?
404
Mark Harrington
@markbernard3rd
When will the love from the left and rage of the right come (just to ultimately flip a few more times)? Thank you for this graph - the haters can see what poll shifts will do this the forecast.
738
Batmantis
@BatMantisRises
I'm a middle of the road person. Followed you awhile. I can tell something is up. I'd like to keep respecting you after the election. ;)
1.4K
Nick Roberts
@Robbyboberts
I'm not a statistician, but if polls shifted to Trump +5 from where they are now, his probability of winning has to be more than what is displayed here. What am I missing?
3.4K
Kurt 🇺🇸(R PA)(6,10,2)☀️ 🌊🏊‍♂️ 🌌🐈⚾️
@MAGAFRV
I mean your basically saying Trump is still the favorite then because he overperformed his polls by 4% in 2020 and 2% or so in 2016. If people really believe he won't do it again, I have a bridge to sell them.
1.4K
Jeremy H. 🦉🏁
@JeremyH418
Just that it got back to 50-50 should be a potent source of embarrassment for MAGA Inc. Then again, most have lost the ability to feel shame.
2.2K
Kurt Schneider
@Cpakurt
Guaranteed pollsters will come out near Election Day saying Trump has it in the bag to discourage Trump voters from going to vote.
492
Brett Goodsman
@Bgoods2711
Nate you’re the best. I like you am an independent leaning libertarian. My question is do you weight the quality of the polls? Not all polls are created equal. For example Morning Consult has little to no value.
450
jacob
@Guardians921
Honestly would not be suprised if she takes off from here and starts to open up a big lead. Her momentum just doesn't seem to be going anywhere
880
Thane
@thanekyrell
The national polling error was 2% and 3% in 16 and 20, both favoring the democrats. Does the model consider historical polling errors at all?
531
patty
@adamsandlerfanx
this chart is kinda deceiving at first glance. if you don't understand what it's trying to do you might draw the conclusion that "oh, if polls are 50/50 then they have an even shot!"
715
jay
@blamelessjay
imagine if people figured out that all of this timewasting gossip actually means nothing at all in real terms. devastating reality crashing down on your weird fantasy that these numbers have any meaning
105
Dogtanian
@Dogtanian139206
Which is another way of saying that your methodology covers all bases, and that come the day of receiving you’ll be able to point to a pile of your ‘predictions’ that proved right while sweeping the others under the carpet. As with top hedge fund managers, bamboozling is the key.
48
Talisman
@Talisman1699853
I do not believe that many people support communism. To support blue is to support communism! I don't believe that many people would support China's way of dealing with it's citizens, there will be no freedom with blue, that has become blatant!
101
Elon Killed this site
@mrgregshields
Again, you gave Trump a 28% chance of winning in 2016, and then claimed you were right and that people just don't understand statistics, so what is the point of any of this? Both sides have between 1-100% chance of winning. Brilliant.
75
Ryan Billy
@RyanG84652613
Dude you use the most inaccurate polls possible. It's so stupid. Additionally, polling does not move as much as these bogus polls indicate.
546
Kalgan 🇺🇦
@hardmeat_brick
Imagine paying $ for this drivel. 😂 Why not explain how it really works, which is that you adjust your model accordingly to get half of the Polymarket money on each side, so Peter Theil can make his 10% juice just like any other bookee?🤔
127
Tea Party Patriot
@Mitchconne75156
I am I reading this right? Kamala has a better chance of winning with a +1 than Trump with +1. 😮‍💨🤔🤔🤔 History would prove otherwise. Trump was out polled by 7 in 2016 and 8 in 2020
3
Dina Claude
@Claudin71829507
You don’t use the most accurate pollsters and then you use the garbage ones . Then you pretend you have a model. You really should be audited because what you are doing is fraudulent.
50
Space Karen
@KMcCrock
What if the GOP NC gov candidate is a Black Nazi who posts on porn sites and Olivia Nuzzi has a thing for whale head?
195
Sky Reckoner
@ReckonerSk68294
I absolutely love how the moment Harris is the infinitesimally mild favorite Nate's comment section is flooded with Republicans whining as opposed to two weeks of Democrat "PETER THIEL" whining.
28
JoshRose
@JoshRos83402777
How confident are you in the reigion of a model that is very sensitive to small changes? At what predicted win probability are you willing to bet on the predicted winner?
25
Fake Felony badge of Honor!
@Grip75001
Nate, you guys all have a problem!
Quote
Nate Cohn
@Nate_Cohn
One question I've seen throughout my replies: does a seemingly strong result for Harris in PA signal a 2020 polling error repeat? That's obviously hard to say, but I want to flag one thing that caught my eye: our polls do not show Harris doing well with the white working class
171
Jack
@GoBills1900
Nate, the 2016 and 2020 elections both saw Trump outperform the swing state polls by between 1 and 7%. Have you factored that in to your analysis?
96
Zeb Shemanski
@ZShemanski35341
this is good look for trump seeing as nates 538 was off by 5 in 2016 and 2020 and if he did not fix it in 2020 no reason to expect he will in 2024
18
freepaolo
@freepaolo1
Does your analysis consider all the fake polls done by paid liberal organizations on behalf of the DNC???
2
John
@JECar1973
Wasn't this the plan all along
@NateSilver538
? You admit you are a raging liberal so your plan of showing Kamala with the lead/"momentum" in your model closer to election makes sense. Atleast be honest next time. We aren't naive
120
Midwit Deathcamp
@countdown2029
People pay for astrology and crypto too. Probability on unique events like this is nothing but numerology. No controlled experiment, no meaning to probability. Ostentatious charlatanism pretending you can tell anything here down to a tenth of a percent! Shameless!
121
BTC
@Bcfromjc
I see that you have stopped pretending she is losing. In a couple of weeks you will start to acknowledge what we all see plainly right now, she is winning.
16
retrobrahhh
@retrobrahhh
what are the odds if you completely disregard morning consult liberal online survey? everyone knows morning consult is fake... even you.
21
Chris Paige
@CPChristianLibe
But this is why your "model" is nonsense - there's just no way that actual voting behavior is this variable. Poli sci is literally based on the idea that voting behavior is based on immutable factors, which means voting behavior is predictable & stable. Granted, there are weak
10