As a retired engineer, who built many models back in the day, I understand why Nate is using all the adjustment factors in the model. However, I think in this case, it’s overcomplicating our analysis. What I learned in my work is that too many adjustment factors can actually lead you astray.
This election is just so ‘unusual’ that I’d be wary of all these adjustments. I prefer to just look at the polls and I haven’t seen many high quality state polls recently. So, all this Electoral College swing recently is much to do about nothing.
Instead, I’d opt for not paying too much attention to what’s happening until Wednesday morning. This will be the debate heard round the world. The state polls that come out 7-10 days after the debate will matter A LOT. After that, it will be a turn out the vote operation like no other.
Yeah, this article strikes me as assuming a 2012/2016 quality and volume of polling. The 2024 polls just aren't good or numerous enough to meaningfully detect a point up or down.
What can we say? Harris has made it close, but isn't running away with it as it looked like she might. The rest is noise.
This comment seems to suggest that Nate started picking out these adjustment factors after the race got going.
That's not the case, though. These adjustment factors were put in place 2, 8, 12 years ago. I remember reading about the convention bounce adjustment in one of Nate's articles way back in 2008! The model's assumptions get double-checked before each cycle (for example, the size of the convention bounce adjustment has shrunk in response to empirical data showing that it is shrinking), but, once the model gets turned on, Nate doesn't go in and fiddle with it. They only go into the model in the first place if it's been shown, empirically, that they are in fact adding meaningful information, rather than just gumming up the works. (You're right that "just gumming up the works" is a common failure state for models of all kinds. I'm a programmer by day! I know!)
Now, in the past, Nate had a big team and the capability to show several different model outputs. He had the Deluxe forecast (which is basically what Silver Bulletin outputs today), as well as the Lite forecast (which was essentially polls-only) and of course the Nowcast (what the Deluxe model says would happen if the election were held today). I miss these options, especially the Nowcast! But the Deluxe forecast is pretty robust and battle-tested at this point, and was usually surprisingly close to the Lite forecast anyway. I think it's wise to trust it.
Building models is a humbling process. Trust me, I remember it well and I was surrounded by many, very intelligent people. What happens is that your model with ‘raw’ data doesn’t quite match reality and you start tinkering with it by adding an ‘adjustment factor’. Pretty soon, you find that you need to add another factor. This process can continue indefinitely, if you are not careful.
But then you get into this uncharted territory, like we are now, where it’s not adding up and you can easily get deluded. We always have to remember that models have limitations and they are ‘trained’ with a certain set of data. If you find yourself in a situation where that past data doesn’t fit the current situation, you need to step back and reevaluate. I’d argue that’s where we are.
I did like your note about the three different levels of past models. Yes, I remember that well and used to look at ones with the rawer data more than the higher-level models for exactly the reasons I am noting here. I am weary of adjustment factors in models having been burned by them in past lives. Been there, done that.
Right, but this is exactly when he SHOULD NOT be tinkering with his past choices, which fit past reality very nicely.
If Nate were sitting up in his tower going, "Hmmm, these numbers don't look right to me. Let's add a convention bounce adjustment," THAT would be disastrous. That's the kind of live, ad-hoc tinkering that not only fatally overcomplicates a model, but biases it.
What actually happened, though, is that Nate added a convention bounce adjustment *16 years ago*. It has worked very nicely ever since. (Even though partisans argued, then as now, that it was a bad modeling choice, the Deluxe model generally turned out to be considerably better calibrated than the Lite model, with less uncertainty and less error.)
Now we have entered uncharted territory, so now is precisely when it would be the WRONG thing for Nate to make an ad hoc edit by removing the long-standing, well-proven convention bounce adjustment. That would be subjecting the model to Nate's biases. Given the relatively small gap between convention-bounce and no-convention-bounce model results, I think it would be much more dangerous for Nate to inject his personal judgment here. We gotta trust the model.
...but, yeah, I do want polls-only and the nowcast back, regardless.
The problem is that previous cycles simply didn't produce the scenario that belied the sloppy way the bounce was implemented. In 2020, Biden got little to no bounce but the model was adjusting expectations down anyway because of Covid. In 2016, Trump and Clinton both got an okay bounce. Unfortunately I can't find an archive for earlier cycles.
The issue isn't the idea of the adjustment. The issue is how it has been implemented. It makes sense to be skeptical of the long-term viability of a bounce following a convention. If a candidate gets a 5-point bump after the convention, there is definitely merit to baking in an assumption that the sugar high won't last and the race will revert to pre-convention numbers eventually. What makes absolutely no sense at all is adjust against a bounce that never happened. Assuming that a lack of bounce somehow magically means a drop of support that for some indescernable reason isn't showing up in the polls (the only explanation being that there actually was a phantom bounce but nobody noticed because it just so happened to coincide with a fundamental drop in support). A lack of a bounce can credibly trim the tail of a candidate's chance of a landslide, but there is no good reason for a lack of a bounce to imply a drop in support.
I think there's a very good reason for a lack of a bounce to imply a drop in support: since most candidates do get a bounce, if you personally have flat averages post-conventions, it at least vigorously suggests that you *did* get a bounce like everybody else but it got cancelled out by some other event.
...and RFK dropped out right there.
You even mention this "phantom bounce" possibility in your comment. You just treat it like a weird off-beat theory that depends on wild coincidence, whereas I think it's empirically the best explanation both for flat post-convention results in general and the Vice-President's flat post-convention results this cycle. I could, of course, turn out to be wrong.
Either way, though, what you're calling for is the opposite of what parent comment was calling for. He was saying that we need fewer adjustments and more simplicity. You're saying that we should have a convention bounce adjustment, but it should have more nuance and complexity to make sure that what it's responding to is really happening in the polls first.
I completley disagree that the 'phantom bounce' is empirically the best theory. I think it's going far too far into reading tea leaves while ignoring the evidence directly in front of us.
Nate has shown in a previous post a clear historical trend towards smaller and smaller convention bounces. 2024 seems to be a continuation of that trend; with Trump there is far too much noise to discern whether he got any bounce or not. There is a very compelling theory of the case that events like conventions simply matter much less than they used to.
What the actual polls have shown over the past few weeks is a very flat race. A slight rise in Harris's odds after the initial surge following the dropout, to a slight dip in the last week or two. Occam's razor applies here: the polls are showing no major change to the race because there has been no major change to the race.
Before RFK dropped out, the polls showed that it was unlikely to help Trump much; it'd maybe give him a fraction of a percent but it looked very unlikely to be a game changer. Then he dropped out. And the polls showed that it wasn't a game changer, helped Trump a little if you squint, just like we thought they would.
But because of an imaginary world that we have absolutely no empirical evidence of existing where Harris got a convention bounce, the model decides to adjust against something that didn't happen.
At the very least, the model could bake in the *possibility*; increase the number of simulations where Harris has a lower voteshare than the polls are showing based on the theory that there is a phantom bounce. But it's not accounting for a (in my opinion very slight) possibility, it is baking it in as a given, based on absolutely no empirical evidence whatsoever. It's complete nonsense in my opinion.
"Either way, though, what you're calling for is the opposite of what parent comment was calling for. "
That's true, I am. If you want simplicity there are other models you can go to for that. Nate's brand is caring about these little details. That said, there is a truth in what OP's saying, that the more you add in these little adjustments, the more complex things get and the more likely you are for bad things to happen. I think this is a clear example of Nate adding a nifty feature that sounds like a good idea, and usually is a good idea, but it has unintended consequences. There are two solutions to that, and there's merit for going in either direction. What there's not merit for is an adjustment that fundamentally doesn't make sense.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
It’s telling that Nate still gets attacked relentlessly for “calling 2016 wrong” because he was the only mainstream pollster showing Trump with a reasonable chance of winning (as he did) while all the forecasters pushing 95-99% chance of a Hillary victory got a pass.
Allan Litchman in 2016 predicted Trump would win when everyone said he was going to lose. Michael Moore also predicted Trump would win despite everyone predicting Trump would lose. Also the other forecasters didn’t predict 95%-99% right before election day. Nate Silver had a 95%-99% forecast for Hillary in the summer of 2016
Allan also posted that Biden was staying in the race and on track to win like five minutes before he dropped out. And the day before 2016 election day pretty much all the mainstream forecasters had her mid-high 90s, Nate was essentially the least bearish on Trump -and he’s never been forgiven for it.
No they didn’t. None of the mainstream forecasters had Hillary’s chance of winning in the 90s. That is false. Nate was just as wrong as all the other ones. Allan Litchman had his reasons for thinking Biden was on track to win. But we won’t ever know since Biden dropped out. Allan got every election correct. Nate screwed up not just in 2016 but also in 2022. Cope
Yep, then he managed to get 2016 and 2022 completely wrong lol. I'll take Allan Litchman getting every presidential election correct since 1984 than Nate Silver who had one good year in predictions which is 2012
TargetSmart appears to be extrapolating the demographic data - age, sex, and race/ethnicity - and then measuring the delta of that extrapolated data. That raises issues of both data quality and significance.
Very small changes in small data sets can produce impressive sounding percent changes. If last year you registered 100 presumed young Latinas and this year you register 150, the 50% increase sounds significant - but it’s only 50 more people, who may or may not really be young Latinas at all.
Lol sounds like you’re just making excuses because the data is actually good data and shows a surge in voter registrations among key demographics. Democrats have seen an increase in voter registration across the country and in key swing states since Kamala Harris became the nominee.
Oh but Nate Silver's forecast is totally not flawed in the way its collected and analyzed. When people register to vote, their age and ethnicity is usually documented.
I'll take Allan Litchman over Nate Silver's flawed forecast
There is a school of thought though that debates don't matter (I don't know if I agree with that myself, just throwing it out there).
The counter argument is that both Trump and Harris are known quantities. What can a debate change in the way of public perceptions of two very well known public figures?
Or maybe its not overrepresentation. Democrats have outperformed polls in 2022 and in 2023/2024 special elections. And Since July 21st, there's been a massive increase in new voter registration in key demographics that are favorable to dems.
I predict Kamala Harris over performs the polls and Nate Silver embarrasses himself again getting it so wrong.
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
And when there’s a 150% increase in new voter registrations among young black women and an 85% increase among all black Americans and a 84% increase among all Latinos. Those new voter registrations aren’t being polled. Therefore, its an underrepresentation. There has been very little polls compared to other election cycles and the most recent polls came from crap pollsters like Patriot Polling and Trafalgar lol
I definitely wouldn't say she's the "opposite" of a known quantity. But the office she's been in is literally the least-meaningful national office, that gives someone the least specific way to define themself. Even though cabinet secretaries have lower profile positions, there are at least specific types of events that they are expected to speak about, and they automatically get some association with the portfolio of that department, whereas the VP doesn't have anything particularly specific.
I'd suggest Joe Biden would argue the point debates don't matter.
And I'd argue this upcoming debate is likely to get more attention than usual. Me, for example. I had littler interest in watching Joe Biden and Trump but Harris/Trump will be must-watch.
Also, Trump was horrible in the last debate but nobody noticed bc Biden was comatose. You also have MSM finally starting to notice that Trump can't speak a single coherent sentence and sounds more like a crazed person on the streets than a candidate for President.
I think there's a good likelihood Trump is the one who comes out of the debate with questions about his fitness for office. Which will be just another reason for anyone who's still undecided to not vote for him.
With such close margins if the debate changes the mind of even half of a percent of voters, it could be enough.
Trump is giving on average an interview a day, and often on platforms that reach out to normies. I suggest your partisan lean is coloring your lens. Most people find his answers to be fine, esp considering neither Harris nor Walz are doing any interviews. They are essentially ceding hundreds of millions of dollars worth of earned media to Trump and Vance.
Interviews with Sean Hannity and Lex Friedman does nothing for Trump; only his base listen to that stuff. "Most people find his answers to be fine" is literally nothing more than a project by you bc I could cite dozens of MSM stories, nightly late night talk shows, dozens of liberal independent media.
The only group it's "fine" with is the red-pilled MAGA crowd.
Trump gave interviews to the national association of black journalists, etc.
From his camp's perspective the content of those interviews is not as important as the fact that he's giving them in the first place. He wants to draw a contrast between himself and Harris but showing up anywhere and everywhere to talk. I expect him to make an appearance at my local 7-11 any day now.
Any seasoned campaign person who advised that content doesn't matter as long as you show up is rather silly.
The news cycle matters at this stage. His appearance at the black journalist talk was a failure in the news cycle and staying on message that wins.
If you show up, stay on message, and give campaign surrogates what they need to go and preach that message you win. Playing defense about calling the opponent not really black stays off the issues that win.
It's a 50/50 race absolutely, but "just showing up is the point" wins zero swing voters if the outcome is swing voters hearing about another silly mistake or reinforcing stereotypes of his.
He's not doing anything that he didn't do in previous elections, he's fine if the goal is hope to win by 10,000 votes in 3 let states. But that's a reason thin margin.
Atleast he is giving interviews and that too long form. . Harris is following the same strategy as Biden by not giving any interviews and it has worked so.far. But I sense public is slowly getting a little frustrated with no interview policy and I think it will backfire
This ‘public’ of which you speak is supposedly worried about a lack of political interviews. Are you imaging tens of millions of Americans laying down their copies of the WSJ and harrumphing to their butlers how unsporting this is! You need to spend more time on planet earth.
Fridman isn't a conservative. He's also done interviews with YouTubers and podcasters who I've never heard of because I'm an old fart. That's why young men are switching bigly to Republicans.
No he's just friends with a lot of conservatives and has soft interviews with conservatives. Young men are not switching bigly to republicans. Democrats and Kamala are still winning young people by double digits
“We want absolutely immaculate water. And we had it. We had H2O. We had the best numbers ever and we were using all the forms of energy, all forms, everything.”
I understand i am not most people but that’s an idiotic statement.
“During my four years, I had the best environmental numbers ever. My top environmental people gave me that statistic just before I walked on the stage, actually.”
That’s also so dumb. You want a fool that talks about a number/statistic whatever that means without anything validating a statement like that? But I guess you do.
That's probably an accurate statement. The real question is why you would take someone seriously when you have to ignore their literal statements because of how idiotic they are? Add that to how 40% of the people he handpicked for his government have come out against him because of how dangerous they deemed him to be during his last term. And that's without considering that by logic there are certain things that you HAVE to take literally... like campaign promises. So he said he would apply a muslim ban and he did. Why should you then interpret differently the fact he said he would have Mexico pay for the wall, and he didn't? Or the fact that he said he would get rid of the deficit and he made it much worse? Those are things normal people would count against a president, but he gets away with it because he says it confidently.
You're going to be so disappointed when Kamala Harris wins like 300+ electoral college votes. New voter registrations in key demographics shows Kamala is going to win big
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
The 'interview' that will matter occurs on Tuesday night. All others pale in comparison. They should be spending their time getting prepared for Tuesday.
Trump has been debating non stop since 2016. Every press conference he held was a debate with a hostile media. This will be a walk in the park for him. Kamala, not so much. Hopefully she's unburdened by what has been.
Like I said, you're going to be so disappointed when Kamala Harris wins like 300+ electoral college votes. New voter registrations in key demographics shows Kamala is going to win big
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
You're going to be so disappointed when Kamala Harris wins like 300+ electoral college votes. New voter registrations in key demographics shows Kamala is going to win big
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
lol you're such a Trump simp. Kamala Harris is going to win the election and will win just as many states Biden won if not more. Dems are going to way over perform the polls just like they did in 2022 and the 2023/2024 special elections. There's been a massive surge in new voter registrations since July 21st in key demographics favorable to dems. These new voter registrations aren't getting factored into the polling
IMO it's very hard to argue debates don't matter this year. You could reasonably argue that debates can't really help a candidate, only hurt - but clearly they have the capacity to matter, even in today's calcified environment.
The general public is only just now getting to know Harris -- partly because they simply don't pay much attention to VPs when they are in office, with a few exceptions, and partly because they remember, unfavorably, her early run in the 2020 cycle. So she has exceptional opportunities to reintroduce herself to the public. Yes, she has been a "public figure," but unless you've been really politically engaged all along, she has not been a big dot on your personal radar screen.
Yes, I have been thinking about this as well. Too many bells/whistles and tinkering, however, reasonable all of the tinkering maybe, it is not clear that the sum of them all will yield the right adjustment. Will the marginal changes in unemployment data or a Fed Rate cut change the outcome? As a maker of many models, the simplest were the best AND their best use was the insights they provided and not the model's actual forecast answer.
To me, it seems like there isn’t nearly enough polling lately. And some of the polling released today is just bizarre if Trump is indeed catching up, like Trump +3 in Texas and +4 in Florida from Emerson. There seems to be less consistency to polling lately, which suggests either a volatile electorate or some major disagreements about the makeup of the electorate.
That’s not the point. The point is that if Texas is +3 Trump, it’s unlikely that PA is tied or +1 Trump. The polls should be somewhat consistent with one another.
Texas is about 5-6 points redder than PA. The two polls in Texas show Trump +4 and +5, and the Pennsylvania polls show Harris +0 or +1, so that sounds about right, right?
The margin between TX and PA in 2020 was a 6.7% difference. The Emerson Poll was +3 Trump in Texas. This doesn’t jibe all that well with a tied race in PA.
Interesting - RCP lists that Emerson Poll as +4 Trump! But still, it's only one poll each for TX and PA in this week, and it's not at all surprising if a true difference of 7 points shows up in one pair of polls as 3 points (or 13 points).
Having family in both places, and having spent a lot of my childhood in one and a good part of my adulthood in the other, I could see that being the case. PA is getting more and more "left behind" and TX is growing more.
Trump appears to have gutted the campaign account to pay his own legal bills, and there probably has also bee siphoning $ by the larger family, so coupled with fund raising issues it’s lead to cutting advertising outside of PA, especially outside the set of swing states that doesn’t include TX.
Without advertising support there it isn’t-that- surprising to find the gap narrower than you’d expect.
You're as delusional as Nate Silver. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
"6% to 3% is not a 50% dropoff - it's a 3% change in the electorate."
The correct way to describe that is 6% to 3% is a RELATIVE 50% Dropoff; but only a 3% ABSOLUTE Dropoff.
"Relative" vs. "Absolute" is especially important when assessing drug claims. Crooked pharmaceutical companies like to shout, "This drug improves survival by 50%!" when the Absolute survival change may only be from 1% to 2%. A 1% improvement is not nothing, but you may choose to take that risk if there are considerably damaging side effects to the drug.
Trump himself could be one explanation. Another is that in every election you have to make way for younger voters, and older ones go to their happy hunting ground.
As an ex-Texan, we all knew that there's a tremendous amount of in migration into the state. Couple that with plenty of deaths of older, conservatives in the last 4 years (about 1 million deaths since 2020) and 2% dropoff would be just about right.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
Isn't the obvious conclusion of this all that the typical convention bounce occurred right after the switch happened, such that when her odds showed 60%, in reality, they were essentially 50%, and now when they show 40%, they're still essentially 50%?
While I get that subjectively tinkering with the model based on what "feels right" without the changes being supported by backtesting leads to worsening performance more often than not, I can't help but feel that, for example, Michigan being 53% odds to Trump doesn't pass the common sense check when looking at the polls.
I'll also add I think while the Traflagar/Patriot/Rasmussens of the world add value in the ensemble model--as they capture some subset of voters the others don't--but when the known to be biased pollsters comprise anything close to the bulk of the model, it opens the possibility of the system going screwy.
This is essentially my view. I've been "de-bouncing" the forecast by looking at Nate's electoral college bias chart, and mapping Kamala's current national poll lead to it. She's currently up by exactly 3, so the Electoral College bias charts would give her chances of winning at somewhere between 56% and 84%.
The RCP polling average is Harris at 1.8. Given what we know about Trump's overperformance in 2016 and 2020 vis a vis the electoral versus popular vote it absolutely seems credible to me that he could lose the popular vote this time out and still win the Presidency based on the EC.
I don't think it really makes a difference. What I focus on is what were the margins at this point of the race in 2016 and 2020 and what were the popular and EC margins versus polling in the actual election.
At this point in both 2016 and 2020 both Clinton and Biden had much larger leads than Clinton currently enjoys according to RCP. I would be surprised if Silver didn't show the same thing. The polling in both races narrowed at the end of the race. Again, I would be surprised if Silver didn't show the same thing. And the final vote margin for Clinton was +2 in the popular vote which meant that she still lost the EC by 56k vote. While Biden won by +4 nationally his EC margin was only 44k votes.
Is it plausible that Trump loses the popular votes by a couple of points this time out but still wins the EC? Absolutely.
Polling underestimated Trump in 2016 and over estimated Biden in 2020. It under estimated Dems in 2022 and dems in special elections in 2023/2024.
New voter registrations since July 21st in key demographics that are helpful to Harris shows that dems are being underestimated in the polls again and that Harris' win will be much bigger than people realize
Everyone seems to want to ignore this important detail
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
Real Clear Politics has Kamala averaging 2.6% which round up its 3%. RCP also excludes many polls and they are partisan. The polls are underestimating Kamala Harris and dems just like they did in 2022
What would have been a "typical" exposure-driven bounce got pulled forward this cycle...so pre-DNC forecast was unadjusted when it "should" have been adjusted, ergo the whipsaw from coin flip+ to coin flip-. The inevitable mean reversion always gets us home.
It's not going to be a coin flip. Kamala Harris is going to over perform the polls by a large margin
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
Another possible conclusion (which doesn't conflict with yours) is that polls are just more accurate this year than they have been in the past. They've figured out the cause of their misses in 2016 and 2020, and now they reflect the true state of the race.
No, polls this year are going to underestimate Kamala Harris and the dems
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
I agree with you the 'bounce' occurred after Harris entered the race. It was baked in by the time the convention came around. I also don't buy Michigan being Trump at this point in time, since few polls indicate that being the case. Once again, my own modeling work years in the past taught me to be careful with all these added 'adjustment' factors. They can get you into hot water fairly quickly.
Nate, you’ve mentioned a few times that betting markets are a sanity check on your model. Does it worry you that your model is starting to diverge from them? Polymarket has Kamala at 50% and predictit has her at 54%
lol wisdom of the crowds. Trump had bigger crowds in 2020 and still lost. But okay, don’t trust the dude who predicted every presidential election correctly since 1984. Go by crowd size lmao
I don’t think she was ever a 5 point favorite (as in, a favorite to win the popular vote by 5 points). She was up to 55-45 or maybe even 60-40 in win probability. But as I said, that is very close to a 40-60 loss probability - only an odds ratio of 2.25 between those two probabilities, which is smaller than the difference between 60-40 and 80-20, or between 80-20 and 90-10.
Let’s be honest with ourselves here, it is not good for trump that a lot of lower information undecided voters are going to hear him formulate thoughts live for the first time in a while. I’m not a coconut pilled hater but it’s gotten bad lately.
I don’t want to veer into partisan trash talk but my view is:
1. The debate will draw many hundreds of thousands of people who primarily get their information from headlines/ clips and have not really tuned in to extended raw footage of either candidate since the last debate.
2. Trump, while not Biden levels of decompensated, is a much harder to follow than he was 8 or even 4 years ago. Less “what a jumbled way of getting to the point” and more “what on earth is he even trying to communicate.”
If it makes you feel better, I think the debate is unlikely to change the state of the race meaningfully. My original response is to the commenter who was projecting that it would, in Trump’s favor.
It really just depends on whether either one ends up looking completely foolish. It doesn’t have to be as bad as Biden’s performance, but if one of them seems comfortable and in charge while the other flails, it could make a difference. Ditto if either one cancels at the last minute. (I’m looking at you, Kamala.)
If they both manage to present a functional facade, then I agree it won’t make much difference.
In any case, it’s something we can all fight about with more data in the debate post-mortem. My forecast: we’ll all be declaring “it’s so over” for the one we don’t like.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
Since Nate took a stab at explaining the “why” behind the relative Trump rise, in association with RFK, and posited he and others may have played it down too much, I’ll ask a question.
Given the RFK voters you’ve met or encountered, how would you measure their engagement in this election?
If you answered anything other than “high”, you’re living on another planet. These people are tuned in. Which means they’re fully aware of RFK’s attempts to get off the ballets in swing states, and they’re keenly aware of how the Left (Democrats) are blocking him. From their perspective, he’s getting screwed again by the same people.
If the Dems had left it alone I suspect most RFK voters would have tuned out. Many would have stayed home on Election Day. Instead, what the Democrats have done is created yet another block of highly disaffected voters, and inadvertently turned them into Trump voters. Enthusiastic ones!
Anything you say that begins with "Tons of RFK supporters" is already missing the point. There aren't tons of RFK voters, and if your feed is showing them to you, it tells you that your feed is not getting anything like a representative sample.
Twitter is a part of real life, but it's only a tiny fraction of real life. Anyone who posts there is highly unrepresentative.
At least for the few RFK voters I know, they are planning on voting for Harris now. A sizable chunk of RFK supporters were really just "double haters" that didn't want Trump or Biden.
Fwiw, my experience is the opposite. I have two friends who were enthusiastic RFK supporters. They are excited for an RFK lead health initiative under Trump.
First, there is only a small minority of people who have ever considered voting for RFK. Out of those people, it's only a tiny fraction who are doing so because they are engaged. That tiny fraction of a small minority may be disproportionately active on Twitter, but they are tiny.
If the D's hadn't done petty crap like deny RFK Secret Service protection he may not have evolved into such a headache. It's bafflingly petty and stupid.
No, matching funding is a very low threshold - only about $100k raised in at least 20 states which almost anyone can meet.
To qualify for secret service protection, a candidate must either poll above 15% for at least 30 consecutive days and be a member of one of the major political parties (a party that won at least 25% of the vote in the last election), or they must poll above 20% for at least 30 consecutive days if they are an independent candidate or minor political party (won between 5% and 25% in the last election). RFK was nowhere near this threshold.
Biden himself directed the Secret Service to give RFK protection, for extra caution. But he definitely was not entitled to it. And he no longer has protection as it ended shortly after he suspended his campaign.
So... I only know one person that I know is an RFK voter, and that person is definitely not highly engaged in this election. They don't watch much news and they don't follow politics, but they have been dabbling in a very confusing swirl of new age philosophy, conspiracy theories, and off the deep end environmentalism. For example, this person believes plants are sentient. While going down internet rabbit holes they stumbled on some of RFKs work and found him to be the only politician they resonated with. I'm certain they are completely unaware of the nuances of DNC or RNC strategies.
Both you and Nate Silver are going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
Nate and Eli, it’s time for your version of a snake graphic. I need to understand the paths to 270 Harris might have opened to go along with this “expected but bad news for Harris” post.
This seems to be a popular line of discussion on the left. From the outside it reads as mild derangement. Please let me know which part of the following you disagree with:
It is our election system. You're saying that our official election system is a challenge for Harris. An intelligent well-informed person saying that our election system is biased demonstrates an intentional disregard for the system by propagating the myth that the popular vote is what matters, when it doesn't. The popular vote is not our system by design. The endless discussions about the "electoral college bias" is reducing election validity in the mind of the audience. I grant you that it's a much milder form of what the right does, but it's really the same thing. You're saying that the election is not valid because it's biased against Harris. But that's not true!
That's not what "bias" means in the world of statistics, and by extension, how Nate is applying it here.
If we want to talk on a societal level, you're underestimating the cynicism-boosting effect of having an electoral system where the candidate the majority of the country wants to win loses in 2 out of the 3 most recent elections. What is there to encourage positive-spirited participation in a system where your vote has zero meaning? Why not just be out for yourself if most people can want one thing, and yet the most important decision is made by people in a handful of states?
Case in point. You're both claiming that the use of "bias" is statistical and following it up with an argument that the Electoral College should be replaced. Reality doesn't line up with preconceived notions all the time, but in this case it's essential to keep hammering on it.
I completely understand the argument for popular vote deciding the presidency. And yet I find the Electoral College to be a good idea. If you can't explain the rationale, then there's more to understand before tearing it down!
The place to start is with what goal you are trying to achieve. The founders' goal was "good governance", and they decided on a non-trivial election mechanism that was not even close to 1-person 1-vote. There are good reasons for republics instead of democracies, and much is written on the topic. Yet no one seems interested in that, just in saying over and over again how it's unfair.
The modern electoral college is nowhere close to what the founders intended though. Representatives to the electoral college were intended to act as a safeguard against an irrational public, as Hamilton articulated in Federalist Paper No. 68. The modern system of binding electors by state law emerged in the 19th century and is nowadays basically just raw mobocracy at the level of the state, instead of the nation. There are some potential reasons to value that, but the modern electoral college cannot be defended in reference to the founder's vision of republicanism.
Yes! That is an important change that was made. It sounds like we both agree that this change probably made the governance worse. My conclusion is that we should NOT take another step on that road (e.g. dispense with EC), but it sounds like yours is the opposite: that because we've already made that change, might as well make more?
If you're asking for my personal view, it's that the modern electoral college system is worse than both pure Republicanism (which benefits from the decision-making of nominal experts) and pure Democracy (which benefits from being impossible to subvert to the whims of narrow interests). Instead, we have a system where the president is chosen by a narrow subset of laymen in a handful of random states, which encourages candidates to promise inane and wasteful patronage to these few individuals (see both Trump and Harris's proposal to eliminate taxes on tips in a bid to win voters in Nevada)
But the Electoral College isn't even remotely still functioning in the role it was intended to play. If we're already voting for presidents directly, which is what the Electoral College was supposed to *prevent*, what is its point?
I prefer it when people on the right simply admit that they want to preserve the Electoral College because it favors them electorally, rather than trying to drape themselves in nonsensical arguments about how it plays an important role in the system when it's already been completely stripped of its intended purpose. That, at least is a defensible position, as both sides are ultimately rooted in personal interest (though I think the Democratic position of "majority vote should win" is far more intuitive, & more in keeping with peer presidential democracies like France).
That's one potential disadvantage for the EC. But look back at history: the Articles of Confederation required a unanimous vote in its version of the national legislature because small states like DE did not want to be dominated by larger states like VA (the CA of its day).
Absent the compromise that the Senate and EC represent what incentive would small states have had for joining the Union? And doesn't that same logic hold true today?
No. States aren't going to secede bc the EC goes away.
From a purely objective viewpoint, the EC distorts the election by making some votes matter more than others. It's hysterical to me that the EC was ostensibly designed so that a small state's votes didn't matter...and what we've ended up doing is making it so that large state's votes don't matter.
The votes of large states obviously do matter, they've just been weighted to address the concerns of small states.
You may think it's unimaginable that states would secede if the EC were abolished, but by the same token isn't it just as unlikely that the country would ever move to abolish the EC? It's just not going to happen.
It's not that I "may think" it's unquestionably true. You think Utah's gonna give up all the benefits of being part of the US bc the EC changes. Never.
So that's a pointless statement.
Regarding the EC...there's already the National Popular Vote compact where states would agree to allocate their electoral votes proportionally based upon the popular vote in their state. This is gaining traction and over time could come to reality. Much, much, much more likely than any state seceding.
The EC was created not to balance small and large states, but to prevent voters from directly choosing Presidents, as the Framers feared that would be dangerous. It no longer functions in this role.
Moreover, the states which are currently benefiting the most from the EC in terms of political attention, such as Pennsylvania, Georgia, and Arizona, are not particularly small states.
If you look at the thread above your message, you can see that other people do not take it in a statistical manner but instead are getting riled up about how the Electoral College is unfair. It demonstrates my point perfectly.
The goal is "good governance", not your limited notion of what's fair.
Is it "fair" that a bum on the street have an equal vote to Kamala Harris herself? Why is that fair? The bum may not even be able to read, know nothing about the economy, know nothing in general. He may be days away from a Fentanyl overdose. Why is it fair that he gets to have a say about the rules that I have a live under?
Why would it be more fair for the populous city centers to unilaterally set rules that rural workmen on the oilfield in the middle of the country have to live under? The votes representing their interest wouldn't matter in that case.
The 2 Electoral College victories in the past 2 decades with a minority of the popular vote gave us the Iraq War and January 6th, and this time might restore a president who attempted to overturn an election with mob violence against the Congress, which was one of the Framers' worst nightmares. That doesn't look like "good governance" to me.
The system is completely fair. The rules are set out ahead of time, and (relatively) impartially adjudicated by the courts.
The electoral college has a purpose, as anyone familiar with history knows. That purpose is to require a presidential candidate to both get a large number of votes and to have a majority in a large number of states.
Regardless of how I feel about the election result, it has served its purpose in recent years.
Don’t mistake undemocratic for unfair. Many things that are less democratic (e.g. an independent judiciary) serve a valuable purpose (said judiciary protecting rights from policies supported by the majority of people).
The electoral college doesn’t require a majority in any state. Look at a three way race (for example, 1992) to see clearly why the electoral college doesn’t do what you claim. The winning candidate only got a majority of the votes in two states that year: Arkansas and DC. The winning candidate received less than 40% in seven states.
*Edit for clarity: Winning candidate above refers to winner of a state’s electoral votes.
Just because everyone knows the rules ahead of time doesn’t mean it is a fair or reasonable system. If we wrote new rules that said coastal states got double the weight of interior states because of how important their port connections are, that could be a fair system by your logic but it would be a terrible system for democracy.
The purpose of the EC, as intended by the Framers, was to prevent voters from directly selecting a President, fearing that that would have a corrupting effect on the country's politics. When I go to vote, I mark a vote directly for a President, not electors. It no longer functions in its intended role.
Nate can defend himself and his verbiage without my commentary... BUT the "bias" is in reference to the popular vote margin not her personally. The Electoral College is completely valid - but in terms of speaking statistically it's fine to use the term "bias" when discussing the correlation between national polling vs EC votes. As a nation we generally speak in terms of national polling ex. "He's winning by 5 points"- so when that does not coincide directly with the EC margin - it is 1) interesting and 2) a strategic campaign topic to discuss
> As a nation we generally speak in terms of national polling
Yes, a focus on the popular vote when that is not the election mechanism and known to be misleading in predictable ways is a choice that someone made. I certainly agree with that. I think that reinforces my point though.
(I agree that it's unfortunate that there isn't a single number that can be easily reported for the balance of Electoral College. It would be much stronger evidence of intentionality if there were and the MSM simply chose not to use it)
> in terms of speaking statistically it's fine to use the term "bias" when discussing the correlation between national polling vs EC votes
So why isn't it the popular vote which is spoken of as biased rather than the EC?
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
There are two elements of how the EC works and which give voters in smaller states disproportionally more power that have nothing to do with its creation in that they weren’t how it always worked and were not the intention. The first is the winner take all proposition. There is nothing in the constitution that says each state should award its electors that way (and a few don’t). Additionally, the framers never contemplated the House (and therefore the EC) having a static size. If the House had continued to grow as it did for the first part of American history, in line with population, the small state bias would be dramatically smaller.
Interesting point about the limited size of the House being an important factor. I hadn’t thought of that. The size-related skew due to the Senate would be much smaller now if the House had grown to have several times as many seats.
I also say that the official election system of Iran is a challenge for reformist candidates. An intelligent well-informed person saying that the Iranian election system is biased demonstrates an intentional disregard for the system by propagating the myth that the preference of the people is what matters, when it doesn't. All that matters, according to the Iranian system, is the vote of the people among the choices approved by the Ayatollah. That is by design. The endless discussions about the bias of the Ayatollahs is pointless moralizing. We shouldn't care what is right when doing politics - all that matters is how the systems are set up.
One doesn't have to think that the Iranian elections are "not valid" to raise objections to how they are carried out, and the same is true for the United States. I don't recommend a revolution in either country - I recommend reform (even though the system is set up in a way that makes particular reforms very difficult to carry out).
Great analogy to the EC arguments being put out here. I think you also hit on the area that probably causes some of the pushback: inability to distinguish between a system that should be improved to be more democratically fair versus cheating or having a “rigged” system. Trump becoming President with fewer votes due the EC disproportionately benefitting him doesn’t mean he cheated.
RE "Why does the Electoral College work this way":
> America’s complex voting system was set up to provide shares of political power, not to simple head counts, but to an intricate network of society’s functions and interests.
> Election of senators by the legislative bodies of the states was intended to tie their representative interests to the states, rather than to national political parties.
> There are many more examples, but the principle is clear: the Constitution is not a poor attempt at equal representation of people, but a complex compromise dividing political power among interests, all of which are required for a functional republic, and which have political needs and interests which they require political power to defend from legal interference.
> So, yes, the "flyover states" are overrepresented if you want America to be a pure 100% democracy, which it never was, and was never intended to be.
> But they are also the reason why you have nice things.
Imagine if the US had never been formed because smaller states in 1787 had refused to join a country where larger states would have dominated the national dialogue.
If you go back and read your Federalist Papers, the EC wasn't just about small state v big state representation. It was also based on the idea that regular people are too stupid to vote for president directly, so they need to elect elite electors instead, who can select a president without being driven by inflamed passions and demagoguery.
We have rightly discarded that as antidemocratic nonsense and now electors are expected to vote for whoever the citizens selected on their ballots, but the EC persists as an outdated relic of 18th-century views.
It's not an either-or proposition. Even if the EC is an anti-populist measure why do small states get a disproportionately larger share of electors? That has everything to do with the small state to large state dichotomy.
And that argument also ignores the issues that would pop up today if the EC and Senate were disposed of. What incentive would smaller states like Utah or Alaska have to participate in that kind of federal union?
If that were really an issue today, I don't think Vermont, Maine, Delaware, Rhode Island, and DC (edit - and Hawaii) would all be part of the National Popular Vote Interstate Compact. Membership in NPVIC correlates strongly with partisanship of a state, but really not at all with size of the state (except insofar as size of the state correlates with partisanship).
On a more serious note, you're correct that small states would never agree to change the system, and that's why it persists. But that's not really relevant to the question of whether the system is fair and democratic. Why is my voice more valuable if I move to Vermont than it is if I move to New York?
It depends on what membership in a country without EC style safeguards would entail. What if CA/NY moved to significantly curtail petroleum drilling? Would AK be okay with that? I doubt it.
200+ years ago, it really was about states wielding political power, and people saw themselves more as citizens of their states than of the US. The system was created to balance Delaware's political power against Virginia's, not to give a citizen of Delaware more votes than a citizen of Virgin. Neither of them really got one anyway.
Go hang out in some oilfield in Texas, the UES in Manhattan, Venice Beach in CA, Provo in UT.
The US is a staggeringly diverse place. In my mind a system that allows for significant state representation is a good thing given that diversity in that it means that only policies that have significant general support ever make it into law.
The static image posted is incorrect, see North Carolina going from D+0.6 to R+0.8 but the change is shown as R+1.6, for example. It looks like it's fixed if you click through to the "interactive chart" but it stoll presents wring in the sustack app and in the emailed version.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
How anyone could call "Trump" an "Uncharismatic Challenger" is beyond me. That makes 5 falses, not four, and very close to the tipping point... especially when people don't necessarily view the short or long term economies as strong, nor do they likely recognize any recent military success.
The dude who lost the popular vote twice and averaged 41% approval for his entire presidency is charismatic? How about you watch the video that gives his reasoning for each one. Charismatic candidates are those with cross over appeal that is well beloved. Trump rambles about Hannibal Lector and is hated by 60% of the country. Also Allan is using the same metrics he has used since 1981. He's not going to change it just because some random person on substack thinks Trump is charismatic. No, its four.
Watch the video, he explains it. Short term and long term economy isn't what peoples public perception is, he's looking just at the raw data. Are we in a rececession? No. Again he's using the same metric that he's used since 1981 to make the predictions. Your opinion on it doesn't matter. Cope
Wrong, Nate's model in 2016 was similar to the other forecast models around the election day. In the summer of 2016, Nate also had a 90-95% chance of Hillary winning like all the other models. Its a myth to claim Nate Silver's forecast was somehow better than the other forecasts. Like I said, I'll take Allan Litchman's predictions since he has gotten every presidential election correct since 1984.
As for judging people by how they vote and the color of their skin. I did no such thing, TargetSmart increase in voter registrations came after July 21st which is when Harris became the nominee and Biden dropped out. Kamala has always done better with black people, hispanics, and women, then Biden. It's obvious that when she became the presumptive nominee on July 21st that the surge in new voter registrations are people that would more likely vote for Kamala. Its not judging anything. Its just a fact, black people and hispanics, especially women, vote for democrats by much larger margins than they do republicans
Lol typical name calling. If what you were saying is true you would provide the evidence but you don't have crap. Cope and seethe. Trumps going to lose. No criminal in the WH for you
Nate, with pro-Trump and anti-Trump presidential juice sucking up all the political forecasting oxygen, there has been next to nothing, here or elsewhere, on Senate and House forecasts. There really needs to be a lot more focus and coverage of that.
For instance, whether or not Harris is able to get over the 270 electoral college hump against Trump, the left's enthusiasm for her over Biden has to have shifted the landscape somewhat on Senate, and particularly House forecasts, including with respect to which party will control each of those chambers, and I would like to see content that analyzes that. Regardless of who is elected President, control of those chambers is going to have a monumental effect on the direction of the country -- e.g., a Trump presidency would be vastly different under Republican control of the Senate and House than Democratic control.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
Cook updated its House race ratings today. Some of it is paywalled, but some interesting analysis there and more interesting charts on Dave Wasserman’s twitter.
Not to cut down on subs, but it's clear a lot of people would be happier if they just wrote "Kamala Harris will win" on a piece of paper, then looked at it every day.
The polls are the polls, they've been wildly off for two Presidential elections straight, particularly for the entire campaign before October (great that they were only off by a point on election day 2016, but the seven months leading up to that had voters being told she was leading by 6-10 points) and for state polls in general (claiming Clinton was up 2 in Ohio when she lost by 8).
The polling data is far better used as a topping for the fundamental data we have about right track/wrong track, voter issues, and the economy. It's idiotic to pretend as if it's likely to be accurate, it's much more likely to not be, and the people with the greatest political literacy all know this.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
Highly unusual circumstances under which Harris gained the nom. Wonder if her convention bounce was after Biden stepped down and now she is just regressing to the mean.
I was really surprised when I saw how dismissive Nate was of the effect of RFK Jr dropping out - both he and the national media seemed to despise the guy to a degree they willfully assumed a candidate with 5% of the vote dropping out wouldn't move the needle. I nearly left a comment on Nate's original coverage of RFK arguing that it was going to have a more significant impact than said, and now I wish I had because I would have been proven right, lol
yeah. I suspect that what happened is 1)Nate + national media weren't expecting this to be a dead heat race, and in previous races that *weren't* close, 3rd party candidates dropping out hasn't historically had a huge impact. So the first mistake was in assuming this wouldn't be a close race (I think they keep having this knee-jerk reaction of expecting elections to go back to pre 2016 patterns) and 2)They assumed that RFK's poorly-run campaign somehow translated to "no-one cares about him", in the face of that literal 5% staring them in the face. You can despise a candidate, but to let that affect your assessment of impact is a very silly thing to do
Its not a close race. Kamala is going to win by a large margin. Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
Do you think that's likely? I would expect more like 2 for not voting, 2 for Trump, 1 for Harris. But RFK was lower in the swing states to begin with than nationally, so this matters, but by less than people think.
Sure, and I may be reading it wrong, but the thrust of this post seems to be “the reason Harris is behind in my forecast is because of bad swing state polls, not because of the convention bounce.” Which isn’t true. If it weren’t for the convention bounce adjustment, she’d be ahead, so people are right to point out that the reason she’s behind in Nate’s forecast is the convention bounce adjustment.
I suppose that’s 1 for not voting, 2 for Trump, 1 for Harris, rather than the 2-2-1 split I was imagining. But I also am not at all convinced that a poll like this accurately measures how many will fail to vote.
lol you posted an article from August the day he dropped out or before and you think that is evidence? And its a DailyMail poll which is not reliable, DailyMail is a right wing tabloid. You can clearly see that RFK didn’t help Trump at all.
Trump lost the election in 2020 by 44k votes in three states. Assume the same general margins this year. What if RFK adds 10k voters to Trump's total in those three states? In my mind that's an extremely modest proposition for somebody who's polling at about 5% nationally.
100%. It was just common sense to assume that RFK WOULD move the needle. there's a difference between him dropping out not having a huge/sizeable impact - no-one expected that a candidate with 5% vote would drive a 10% swing or any - versus having not even a 1% impact, which the media projected for no reason other than...wishful thinking on their part I assume
For a 5% candidate dropping out to have a 1% impact, you need a very strong effect. A substantial fraction of those people weren't actually going to vote to begin with (that is always the case with people who poll as supporting a third party candidate) and once he's dropped out they are even less likely to vote. You need a very lopsided break for a fraction of 5% to make a 1% difference on net.
That's 5 million total across all 50 states. About 20% of voters are in the swing states, so that's about 1 million total across all 7 states, or about 200,000 in any one big swing state (assuming that swing state voters are just as likely to vote third party as voters elsewhere, which seems like a bad assumption).
Getting a margin of 30,000 out of 200,000 requires something close to a 60-40 split, with no abstentions. It's not out of the realm of possibility, but not particularly likely.
60-40 is within the realm of possibility just in terms of the "natural" voter split for RFK supporters. Now add in an endorsement? Definitely plausible
You’re talking about AZ (~11k), GA (~11k), & WI (~20k). It’s plausible for Trump to flip all three with RFK Jr endorsement but Harris is polling pretty well (+3.2 avg per 270toWin) in WI, and Kennedy in WI seemed more like 4% avg. before he dropped out.
Good point. Man oh man the 2020 WI polls underestimated Trump so badly. Like by ~7%. On the other hand, 2022 WI Senate polls were pretty accurate, although they did underestimate the very weak defunder D candidate Barnes, who ended up losing by only 27k even though polling average was -3.4%.
Trump lost PA by 80,550 votes. Lost Wisconsin by 20,682 votes, and lost Michigan by 154,188 votes. Even if RFK adds 10,000 voters for Trump in each state. Trump would still lose
RFK dropping out didn't have any real impact. Kamala Harris was still seeing increase leads even after RFK dropped out. Its only been tightening within this last week.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
On June 26, your Biden vs Trump model had Biden's odds of victory at 33.7%. You wrote, "As our model launches, either Biden or Trump could easily win — but the odds are in the ex-president’s favor." The headline was "the presidential election isn't a toss-up."
Today's Biden vs. Harris model has Harris' odds of victory at 38.3%. Given the similar odds of victory for Biden on June 26 as Harris on September 6, could you write the same headline you wrote about Biden about today's race?
"Either Harris or Trump could easily win — but the odds are in the ex-president’s favor. The presidential election isn't a toss-up."
I have been thinking the same thing, but I think Nate is waiting for more state polling to confirm [PA]. I'm wondering if 2 or 3 decent Kamala polls by the right pollsters would wildly swing the forecast.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
I think Nate is constantly thinking about how the model works and what information it doesn't have. He has been clear that he doesn't like to make adjustments to the model mid-stream, but that hasn't stopped him from highlighting when he think the model may be temporarily biased in one direction or the other. For example, after the debate Nate was clear that he thought Biden's position in the model was overly optimistic.
You're talking about June 26, which was one day before the debate, but even at that point I think Nate had a sour enough view of Biden that he might have been applying some mental downward pressure to Biden's chances. I mean, Nate was saying Biden should drop out long before most people.
Also, 33.7% is not the same as 38.3%. You have to draw the line from "toss-up" to "not-a-toss-up" somewhere, and where you draw that line may very well be between those two numbers. 35% for example. I mean, maybe it's 40% and the race has moved into non-toss-up category, but if I were Nate I'd wait Harris has been under 40% for a week before I declared "not a toss up"
I think the choice to apply a penalty to adjust for a postconvention bounce is a defensible one, I just think it’s open question how applicable it is to this particular election. We’ll find out soon.
Yeah I think that's a very fair point, and is probably another reason Nate isn't declaring it "not a tossup." He's stated many times that if Kamala's current polling lead holds her win% in the model will get better over time.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
It’s not stubbornness. It’s process. If you adjust the model midstream as you see different things happen but before getting any actual results, you aren’t modeling based on statistics, you’re being a pundit. Nate sets his own rules at the start of the process, that unless there is something that is clearly and obviously wrong/broken, he doesn’t adjust the model midstream, he makes adjustments between elections. In this case it is highly possible that this is leading the model to be off at the moment ( although if so it will wash out long before the election) but saying he isn’t making an adjustment because of stubbornness is just a really strange perspective.
Nope. Its stubbornness. Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
I guess you're not taking into account what we saw in the Kansas reference after Dobbs - that enthusiasm among women - of all ages, but especially 18-29 - is very high and they may come out in numbers larger than any poll is predicting. Turnout is very hard to predict. Actually I wouldn't call it "enthusiasm". I would call it female "RAGE". Donald Trump offers us nothing. And we have everythihng to lose with Trump and the MAGA GOP. So there's that. Put that in your model and smoke it.
And how's your sugar-daddy Peter Thiel doing? Got anything bet on the race yourself, Nate?
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
Your current state polling averages have Harris narrowly ahead in PA, WI, MI, GA, and NV. But your forecasts have Trump narrowly more likely to win all of these states, and therefore more likely to win the Electoral College. The inescapable conclusion is that your model assumes we are about to see a shift towards Trump in PA, WI, MI, GA, and NV (as well as nationally). Right?
It's unclear to me how much of that assumption in the model — that things are about to shift towards Trump — is due to the convention bounce adjustment. Are there reasons other than the convention bounce adjustment that the model assumes things are about to shift towards Trump?
If not, if that assumption is 100% due to the convention bounce adjustment, then am I correct in thinking that the model would have Harris favored to win if all the polling averages are exactly the same in 2 weeks as they are today?
I believe yes, that's what Nate says about the adjustment towards Trump "it will begin to work its way out of the model over the course of the next week or two"
Imagine, hypothetically, that 2 weeks from today, state and national polling averages have moved towards Harris by just 1 percentage point relative to where they are today. In this hypothetical scenario, her win probability in your model will have spiked — from under 40% today to well above 50% in 2 weeks. Right?
Won't people be confused as to why that is happening, if you aren't clear now that her current win probability is being lowered by the convention bounce assumption?
From your point of view, as someone who has to explain what the model is doing to everyday readers, it seems to me like it's in your strategic interest to be clear and open about the effect of the convention bounce assumption right now!
It seems to me that if you ran a version of your model today with NO convention bounce assumption, it would show Harris as the narrow favorite to win. Am I right in thinking that? Why would it be a problem for you to admit that that is the case, while still saying that you think the convention bounce assumption is the right assumption?
Sure. I guess I am responding partly to Nate's saltiness on Twitter, where he has been very annoyed by people correctly pointing out that the only reason his model has Harris behind right now is this convention bounce assumption.
I think it would be helpful if he clearly stated the obvious, which is that his model would have Harris as the favorite to win right now if it weren't for that one specific assumption – which is in his view good and defensible assumption. And which, in his defense, is an assumption he has built into the model every 4 years for as far back as I can remember, so it's not like it's something he's added in capriciously or for any nefarious reasons.
Instead, he seems very determined to point out problems for Harris that are NOT the convention bounce assumption. I’m not denying those problems are real – of course it’s bad that her leads have slightly declined in state polling. But when your model is a bit of a bullish-for-Trump outlier among models, as Nate’s is right now, it seems disingenuous to focus on reasons for that bullishness that are anything other than “the model assumes Harris is in the middle of a convention bounce that is about to fade.” Because if the model didn’t assume that 1 specific thing, it wouldn’t have Trump favored to win.
He's stated in multiple articles here that Trump's current lead in the forecast is due to the model applying a DNC convention boost. I haven't read his X posts about it, but from what I've seen here, he has been very clear about that.
I remember an article here recently saying that if the polling today were the same the day before the election, the model would have Harris as a clear favorite. I'm not sure how much clearer he can be.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
My question is are all the people, especially boomers who died since 2020 and the surge in recent registrations among Dem leaning voters factored into the PA and NC numbers? I admit that I am severely uneducated in statistics and modeling.
As I understand it, the forecast assumes that things like changes in the electorate and enthusiasm levels will ultimately be reflected in polls. Lots of pollsters already weight responses to account for demographic shifts, so you wouldn’t want a model to double down and over-skew as a forecaster.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
As a retired engineer, who built many models back in the day, I understand why Nate is using all the adjustment factors in the model. However, I think in this case, it’s overcomplicating our analysis. What I learned in my work is that too many adjustment factors can actually lead you astray.
This election is just so ‘unusual’ that I’d be wary of all these adjustments. I prefer to just look at the polls and I haven’t seen many high quality state polls recently. So, all this Electoral College swing recently is much to do about nothing.
Instead, I’d opt for not paying too much attention to what’s happening until Wednesday morning. This will be the debate heard round the world. The state polls that come out 7-10 days after the debate will matter A LOT. After that, it will be a turn out the vote operation like no other.
Yeah, this article strikes me as assuming a 2012/2016 quality and volume of polling. The 2024 polls just aren't good or numerous enough to meaningfully detect a point up or down.
What can we say? Harris has made it close, but isn't running away with it as it looked like she might. The rest is noise.
"Harris has made it close, but isn't running away with it as it looked like she might. The rest is noise."
In other words, she might be trailing *or* running away in a blue wave and we won't know until we have solid post-debate polling.
This comment seems to suggest that Nate started picking out these adjustment factors after the race got going.
That's not the case, though. These adjustment factors were put in place 2, 8, 12 years ago. I remember reading about the convention bounce adjustment in one of Nate's articles way back in 2008! The model's assumptions get double-checked before each cycle (for example, the size of the convention bounce adjustment has shrunk in response to empirical data showing that it is shrinking), but, once the model gets turned on, Nate doesn't go in and fiddle with it. They only go into the model in the first place if it's been shown, empirically, that they are in fact adding meaningful information, rather than just gumming up the works. (You're right that "just gumming up the works" is a common failure state for models of all kinds. I'm a programmer by day! I know!)
Now, in the past, Nate had a big team and the capability to show several different model outputs. He had the Deluxe forecast (which is basically what Silver Bulletin outputs today), as well as the Lite forecast (which was essentially polls-only) and of course the Nowcast (what the Deluxe model says would happen if the election were held today). I miss these options, especially the Nowcast! But the Deluxe forecast is pretty robust and battle-tested at this point, and was usually surprisingly close to the Lite forecast anyway. I think it's wise to trust it.
Building models is a humbling process. Trust me, I remember it well and I was surrounded by many, very intelligent people. What happens is that your model with ‘raw’ data doesn’t quite match reality and you start tinkering with it by adding an ‘adjustment factor’. Pretty soon, you find that you need to add another factor. This process can continue indefinitely, if you are not careful.
But then you get into this uncharted territory, like we are now, where it’s not adding up and you can easily get deluded. We always have to remember that models have limitations and they are ‘trained’ with a certain set of data. If you find yourself in a situation where that past data doesn’t fit the current situation, you need to step back and reevaluate. I’d argue that’s where we are.
I did like your note about the three different levels of past models. Yes, I remember that well and used to look at ones with the rawer data more than the higher-level models for exactly the reasons I am noting here. I am weary of adjustment factors in models having been burned by them in past lives. Been there, done that.
Right, but this is exactly when he SHOULD NOT be tinkering with his past choices, which fit past reality very nicely.
If Nate were sitting up in his tower going, "Hmmm, these numbers don't look right to me. Let's add a convention bounce adjustment," THAT would be disastrous. That's the kind of live, ad-hoc tinkering that not only fatally overcomplicates a model, but biases it.
What actually happened, though, is that Nate added a convention bounce adjustment *16 years ago*. It has worked very nicely ever since. (Even though partisans argued, then as now, that it was a bad modeling choice, the Deluxe model generally turned out to be considerably better calibrated than the Lite model, with less uncertainty and less error.)
Now we have entered uncharted territory, so now is precisely when it would be the WRONG thing for Nate to make an ad hoc edit by removing the long-standing, well-proven convention bounce adjustment. That would be subjecting the model to Nate's biases. Given the relatively small gap between convention-bounce and no-convention-bounce model results, I think it would be much more dangerous for Nate to inject his personal judgment here. We gotta trust the model.
...but, yeah, I do want polls-only and the nowcast back, regardless.
Agreed, it would be dangerous to tinker with your system due to some narrative in your head about this year being unusual
The problem is that previous cycles simply didn't produce the scenario that belied the sloppy way the bounce was implemented. In 2020, Biden got little to no bounce but the model was adjusting expectations down anyway because of Covid. In 2016, Trump and Clinton both got an okay bounce. Unfortunately I can't find an archive for earlier cycles.
The issue isn't the idea of the adjustment. The issue is how it has been implemented. It makes sense to be skeptical of the long-term viability of a bounce following a convention. If a candidate gets a 5-point bump after the convention, there is definitely merit to baking in an assumption that the sugar high won't last and the race will revert to pre-convention numbers eventually. What makes absolutely no sense at all is adjust against a bounce that never happened. Assuming that a lack of bounce somehow magically means a drop of support that for some indescernable reason isn't showing up in the polls (the only explanation being that there actually was a phantom bounce but nobody noticed because it just so happened to coincide with a fundamental drop in support). A lack of a bounce can credibly trim the tail of a candidate's chance of a landslide, but there is no good reason for a lack of a bounce to imply a drop in support.
I think there's a very good reason for a lack of a bounce to imply a drop in support: since most candidates do get a bounce, if you personally have flat averages post-conventions, it at least vigorously suggests that you *did* get a bounce like everybody else but it got cancelled out by some other event.
...and RFK dropped out right there.
You even mention this "phantom bounce" possibility in your comment. You just treat it like a weird off-beat theory that depends on wild coincidence, whereas I think it's empirically the best explanation both for flat post-convention results in general and the Vice-President's flat post-convention results this cycle. I could, of course, turn out to be wrong.
Either way, though, what you're calling for is the opposite of what parent comment was calling for. He was saying that we need fewer adjustments and more simplicity. You're saying that we should have a convention bounce adjustment, but it should have more nuance and complexity to make sure that what it's responding to is really happening in the polls first.
I completley disagree that the 'phantom bounce' is empirically the best theory. I think it's going far too far into reading tea leaves while ignoring the evidence directly in front of us.
Nate has shown in a previous post a clear historical trend towards smaller and smaller convention bounces. 2024 seems to be a continuation of that trend; with Trump there is far too much noise to discern whether he got any bounce or not. There is a very compelling theory of the case that events like conventions simply matter much less than they used to.
What the actual polls have shown over the past few weeks is a very flat race. A slight rise in Harris's odds after the initial surge following the dropout, to a slight dip in the last week or two. Occam's razor applies here: the polls are showing no major change to the race because there has been no major change to the race.
Before RFK dropped out, the polls showed that it was unlikely to help Trump much; it'd maybe give him a fraction of a percent but it looked very unlikely to be a game changer. Then he dropped out. And the polls showed that it wasn't a game changer, helped Trump a little if you squint, just like we thought they would.
But because of an imaginary world that we have absolutely no empirical evidence of existing where Harris got a convention bounce, the model decides to adjust against something that didn't happen.
At the very least, the model could bake in the *possibility*; increase the number of simulations where Harris has a lower voteshare than the polls are showing based on the theory that there is a phantom bounce. But it's not accounting for a (in my opinion very slight) possibility, it is baking it in as a given, based on absolutely no empirical evidence whatsoever. It's complete nonsense in my opinion.
"Either way, though, what you're calling for is the opposite of what parent comment was calling for. "
That's true, I am. If you want simplicity there are other models you can go to for that. Nate's brand is caring about these little details. That said, there is a truth in what OP's saying, that the more you add in these little adjustments, the more complex things get and the more likely you are for bad things to happen. I think this is a clear example of Nate adding a nifty feature that sounds like a good idea, and usually is a good idea, but it has unintended consequences. There are two solutions to that, and there's merit for going in either direction. What there's not merit for is an adjustment that fundamentally doesn't make sense.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
It’s telling that Nate still gets attacked relentlessly for “calling 2016 wrong” because he was the only mainstream pollster showing Trump with a reasonable chance of winning (as he did) while all the forecasters pushing 95-99% chance of a Hillary victory got a pass.
Allan Litchman in 2016 predicted Trump would win when everyone said he was going to lose. Michael Moore also predicted Trump would win despite everyone predicting Trump would lose. Also the other forecasters didn’t predict 95%-99% right before election day. Nate Silver had a 95%-99% forecast for Hillary in the summer of 2016
Allan also posted that Biden was staying in the race and on track to win like five minutes before he dropped out. And the day before 2016 election day pretty much all the mainstream forecasters had her mid-high 90s, Nate was essentially the least bearish on Trump -and he’s never been forgiven for it.
No they didn’t. None of the mainstream forecasters had Hillary’s chance of winning in the 90s. That is false. Nate was just as wrong as all the other ones. Allan Litchman had his reasons for thinking Biden was on track to win. But we won’t ever know since Biden dropped out. Allan got every election correct. Nate screwed up not just in 2016 but also in 2022. Cope
I'm just give you this short history of the 2016 election and move right along. Cheerio.
Nate Silver's Finest Hour: https://goodreason.substack.com/p/nate-silvers-finest-hour-part-1-of
Yep, then he managed to get 2016 and 2022 completely wrong lol. I'll take Allan Litchman getting every presidential election correct since 1984 than Nate Silver who had one good year in predictions which is 2012
https://www.youtube.com/watch?v=JoWt1EOA340
Sigh. You didn't read it, did you?
I find ppl’s profound misunderstanding of basic probabilities in forecasting to be fascinating.
TargetSmart appears to be extrapolating the demographic data - age, sex, and race/ethnicity - and then measuring the delta of that extrapolated data. That raises issues of both data quality and significance.
Very small changes in small data sets can produce impressive sounding percent changes. If last year you registered 100 presumed young Latinas and this year you register 150, the 50% increase sounds significant - but it’s only 50 more people, who may or may not really be young Latinas at all.
Lol sounds like you’re just making excuses because the data is actually good data and shows a surge in voter registrations among key demographics. Democrats have seen an increase in voter registration across the country and in key swing states since Kamala Harris became the nominee.
https://www.nytimes.com/2024/08/13/upshot/democrats-voter-registration.html
Nate Silver is the only person who is forecasting Kamala losing. It’s absurd. He’s wrong.
No, I am pointing out shortcomings of a specific data set that the people who collected it actually acknowledge.
Increases in people registering as Democrats are verifiable. Specific claims about voter demographic details are interpolation, not verifiable data.
Currently pretty much everyone considers this a toss up, including Nate. Complacency is not something to encourage.
Oh but Nate Silver's forecast is totally not flawed in the way its collected and analyzed. When people register to vote, their age and ethnicity is usually documented.
I'll take Allan Litchman over Nate Silver's flawed forecast
https://www.youtube.com/watch?v=JoWt1EOA340
Long live the Nowcast.
There is a school of thought though that debates don't matter (I don't know if I agree with that myself, just throwing it out there).
The counter argument is that both Trump and Harris are known quantities. What can a debate change in the way of public perceptions of two very well known public figures?
Harris is the opposite of a known quantity, that why her numbers are so volatile
That's one possibility. The other is that D enthusiasm in the aftermath of the Biden resignation resulted in overrepresentation in polling.
The two are not mutually exclusive. She definitely got a bump from the Biden resignation and also nobody really knows much about her.
Or maybe its not overrepresentation. Democrats have outperformed polls in 2022 and in 2023/2024 special elections. And Since July 21st, there's been a massive increase in new voter registration in key demographics that are favorable to dems.
I predict Kamala Harris over performs the polls and Nate Silver embarrasses himself again getting it so wrong.
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
If Harris' poll numbers start to drop and Trump makes up ground it's overrepresentation. It's called a "bump" for a reason.
And when there’s a 150% increase in new voter registrations among young black women and an 85% increase among all black Americans and a 84% increase among all Latinos. Those new voter registrations aren’t being polled. Therefore, its an underrepresentation. There has been very little polls compared to other election cycles and the most recent polls came from crap pollsters like Patriot Polling and Trafalgar lol
I definitely wouldn't say she's the "opposite" of a known quantity. But the office she's been in is literally the least-meaningful national office, that gives someone the least specific way to define themself. Even though cabinet secretaries have lower profile positions, there are at least specific types of events that they are expected to speak about, and they automatically get some association with the portfolio of that department, whereas the VP doesn't have anything particularly specific.
I'd suggest Joe Biden would argue the point debates don't matter.
And I'd argue this upcoming debate is likely to get more attention than usual. Me, for example. I had littler interest in watching Joe Biden and Trump but Harris/Trump will be must-watch.
Also, Trump was horrible in the last debate but nobody noticed bc Biden was comatose. You also have MSM finally starting to notice that Trump can't speak a single coherent sentence and sounds more like a crazed person on the streets than a candidate for President.
I think there's a good likelihood Trump is the one who comes out of the debate with questions about his fitness for office. Which will be just another reason for anyone who's still undecided to not vote for him.
With such close margins if the debate changes the mind of even half of a percent of voters, it could be enough.
Trump is giving on average an interview a day, and often on platforms that reach out to normies. I suggest your partisan lean is coloring your lens. Most people find his answers to be fine, esp considering neither Harris nor Walz are doing any interviews. They are essentially ceding hundreds of millions of dollars worth of earned media to Trump and Vance.
Interviews with Sean Hannity and Lex Friedman does nothing for Trump; only his base listen to that stuff. "Most people find his answers to be fine" is literally nothing more than a project by you bc I could cite dozens of MSM stories, nightly late night talk shows, dozens of liberal independent media.
The only group it's "fine" with is the red-pilled MAGA crowd.
Trump gave interviews to the national association of black journalists, etc.
From his camp's perspective the content of those interviews is not as important as the fact that he's giving them in the first place. He wants to draw a contrast between himself and Harris but showing up anywhere and everywhere to talk. I expect him to make an appearance at my local 7-11 any day now.
Any seasoned campaign person who advised that content doesn't matter as long as you show up is rather silly.
The news cycle matters at this stage. His appearance at the black journalist talk was a failure in the news cycle and staying on message that wins.
If you show up, stay on message, and give campaign surrogates what they need to go and preach that message you win. Playing defense about calling the opponent not really black stays off the issues that win.
It's a 50/50 race absolutely, but "just showing up is the point" wins zero swing voters if the outcome is swing voters hearing about another silly mistake or reinforcing stereotypes of his.
He's not doing anything that he didn't do in previous elections, he's fine if the goal is hope to win by 10,000 votes in 3 let states. But that's a reason thin margin.
Yea he spoke there and it was completely embarrassing, he looked awful and his favor abilities dropped
I don't think Lex Fridman is popular with any element of Trump's base.
White men aren't part of Trump's base? I'd hazard to guess that's the majority of Fridman's audience
Is he popular with anyone?
Atleast he is giving interviews and that too long form. . Harris is following the same strategy as Biden by not giving any interviews and it has worked so.far. But I sense public is slowly getting a little frustrated with no interview policy and I think it will backfire
This ‘public’ of which you speak is supposedly worried about a lack of political interviews. Are you imaging tens of millions of Americans laying down their copies of the WSJ and harrumphing to their butlers how unsporting this is! You need to spend more time on planet earth.
Tuesday night is what matters.
Biden did that and he won lol.
Fridman isn't a conservative. He's also done interviews with YouTubers and podcasters who I've never heard of because I'm an old fart. That's why young men are switching bigly to Republicans.
No he's just friends with a lot of conservatives and has soft interviews with conservatives. Young men are not switching bigly to republicans. Democrats and Kamala are still winning young people by double digits
It also gives Trump a chance to eff up in front of a friendly crowd, and then turned into a Dem campaign ad.
“We want absolutely immaculate water. And we had it. We had H2O. We had the best numbers ever and we were using all the forms of energy, all forms, everything.”
I understand i am not most people but that’s an idiotic statement.
“During my four years, I had the best environmental numbers ever. My top environmental people gave me that statistic just before I walked on the stage, actually.”
That’s also so dumb. You want a fool that talks about a number/statistic whatever that means without anything validating a statement like that? But I guess you do.
Trump haters take Trump literally but not seriously. Trump supporters take Trump seriously but not literally.
That's probably an accurate statement. The real question is why you would take someone seriously when you have to ignore their literal statements because of how idiotic they are? Add that to how 40% of the people he handpicked for his government have come out against him because of how dangerous they deemed him to be during his last term. And that's without considering that by logic there are certain things that you HAVE to take literally... like campaign promises. So he said he would apply a muslim ban and he did. Why should you then interpret differently the fact he said he would have Mexico pay for the wall, and he didn't? Or the fact that he said he would get rid of the deficit and he made it much worse? Those are things normal people would count against a president, but he gets away with it because he says it confidently.
You're going to be so disappointed when Kamala Harris wins like 300+ electoral college votes. New voter registrations in key demographics shows Kamala is going to win big
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
The 'interview' that will matter occurs on Tuesday night. All others pale in comparison. They should be spending their time getting prepared for Tuesday.
Trump has been debating non stop since 2016. Every press conference he held was a debate with a hostile media. This will be a walk in the park for him. Kamala, not so much. Hopefully she's unburdened by what has been.
Like I said, you're going to be so disappointed when Kamala Harris wins like 300+ electoral college votes. New voter registrations in key demographics shows Kamala is going to win big
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
Yes, "Trump is giving on average an interview a day." How many interviews is Kamala giving?
You're going to be so disappointed when Kamala Harris wins like 300+ electoral college votes. New voter registrations in key demographics shows Kamala is going to win big
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
Vance blew it today. Tucker Carlson had a long interview with a Holocaust “revisionist” on X. Guy says Churchill was the bad guy in WW2.
When asked, JD Vance refused to disavow Carlson. Vance was a bad pick.
lol you're such a Trump simp. Kamala Harris is going to win the election and will win just as many states Biden won if not more. Dems are going to way over perform the polls just like they did in 2022 and the 2023/2024 special elections. There's been a massive surge in new voter registrations since July 21st in key demographics favorable to dems. These new voter registrations aren't getting factored into the polling
"I'd suggest Joe Biden would argue the point debates don't matter." -- you win the comments section.
Biden is clearly the exception to the rule, but that is an extreme set of circumstances.
So far as public interest goes, weren't the debate ratings substantially lower than in 2020?
Yes. But based on enthusiasm from both sides I would suggest Tuesday's vote will
1. Have a much bigger audience than recent debates including a lot of younger people who had zero interest in watching two people born before 1950
2. Has the potential to sway more voters than usual (still a small number overall, but a significant number in terms of election outcome).
84 million watched the first Clinton-Trump debate.
73 million watched the first Biden-Trump debate.
51 million watched the 2024 Biden-Trump debate.
50-60 million sounds plausible for Harris-Trump.
IMO it's very hard to argue debates don't matter this year. You could reasonably argue that debates can't really help a candidate, only hurt - but clearly they have the capacity to matter, even in today's calcified environment.
"Argument that debates don't matter"
The debate sure seemed to matter to Biden, his family, Kamala, and the DNC.
The general public is only just now getting to know Harris -- partly because they simply don't pay much attention to VPs when they are in office, with a few exceptions, and partly because they remember, unfavorably, her early run in the 2020 cycle. So she has exceptional opportunities to reintroduce herself to the public. Yes, she has been a "public figure," but unless you've been really politically engaged all along, she has not been a big dot on your personal radar screen.
My view is Harris is not a known quantity at all.
Yes, I have been thinking about this as well. Too many bells/whistles and tinkering, however, reasonable all of the tinkering maybe, it is not clear that the sum of them all will yield the right adjustment. Will the marginal changes in unemployment data or a Fed Rate cut change the outcome? As a maker of many models, the simplest were the best AND their best use was the insights they provided and not the model's actual forecast answer.
Harris hired Obama‘s team and I’m glad she did. They had a fantastic ground game both times.
To me, it seems like there isn’t nearly enough polling lately. And some of the polling released today is just bizarre if Trump is indeed catching up, like Trump +3 in Texas and +4 in Florida from Emerson. There seems to be less consistency to polling lately, which suggests either a volatile electorate or some major disagreements about the makeup of the electorate.
Anything that shows a close race in either TX or FL makes me think hard about how trustworthy polling is right now.
Texas has been a close but lean R state for a while there. Close polls there would be consistent with 2020 and 2022 election results.
That’s not the point. The point is that if Texas is +3 Trump, it’s unlikely that PA is tied or +1 Trump. The polls should be somewhat consistent with one another.
Texas is about 5-6 points redder than PA. The two polls in Texas show Trump +4 and +5, and the Pennsylvania polls show Harris +0 or +1, so that sounds about right, right?
The margin between TX and PA in 2020 was a 6.7% difference. The Emerson Poll was +3 Trump in Texas. This doesn’t jibe all that well with a tied race in PA.
Interesting - RCP lists that Emerson Poll as +4 Trump! But still, it's only one poll each for TX and PA in this week, and it's not at all surprising if a true difference of 7 points shows up in one pair of polls as 3 points (or 13 points).
Texas and PA are converging by a couple points a cycle
Having family in both places, and having spent a lot of my childhood in one and a good part of my adulthood in the other, I could see that being the case. PA is getting more and more "left behind" and TX is growing more.
Trump appears to have gutted the campaign account to pay his own legal bills, and there probably has also bee siphoning $ by the larger family, so coupled with fund raising issues it’s lead to cutting advertising outside of PA, especially outside the set of swing states that doesn’t include TX.
Without advertising support there it isn’t-that- surprising to find the gap narrower than you’d expect.
Trump will win both states by close to 10. It won’t even be competitive.
Red-pilled Trumper has entered the chat.
So let's bet. Name your wager. Guessing you won't want to put your money where your Internet comment is.
Stop committing crimes in public. You.
He’s not gonna win Pennsylvania by 10, that’s nonsense. Probably not Texas either. Harris isn’t Carter and Trump isn’t Reagan.
$1000.00 Trump loses
You're as delusional as Nate Silver. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
So let's bet. Name your wager.
$1000.00 Kamala wins
TX was +6 Trump in 2020. A three point margin is plausible I suppose but I would be curious as to the explanation for a 50% drop off.
6% to 3% is not a 50% dropoff - it's a 3% change in the electorate.
"6% to 3% is not a 50% dropoff - it's a 3% change in the electorate."
The correct way to describe that is 6% to 3% is a RELATIVE 50% Dropoff; but only a 3% ABSOLUTE Dropoff.
"Relative" vs. "Absolute" is especially important when assessing drug claims. Crooked pharmaceutical companies like to shout, "This drug improves survival by 50%!" when the Absolute survival change may only be from 1% to 2%. A 1% improvement is not nothing, but you may choose to take that risk if there are considerably damaging side effects to the drug.
What's the explanation?
Texas has been trending blue for years.
Trump himself could be one explanation. Another is that in every election you have to make way for younger voters, and older ones go to their happy hunting ground.
As an ex-Texan, we all knew that there's a tremendous amount of in migration into the state. Couple that with plenty of deaths of older, conservatives in the last 4 years (about 1 million deaths since 2020) and 2% dropoff would be just about right.
We can expect polls and results to get bluer as time goes by. People die off, people come of age.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
Florida too in 2020. 2022 was anomalous for Florida.
Isn't the obvious conclusion of this all that the typical convention bounce occurred right after the switch happened, such that when her odds showed 60%, in reality, they were essentially 50%, and now when they show 40%, they're still essentially 50%?
While I get that subjectively tinkering with the model based on what "feels right" without the changes being supported by backtesting leads to worsening performance more often than not, I can't help but feel that, for example, Michigan being 53% odds to Trump doesn't pass the common sense check when looking at the polls.
I'll also add I think while the Traflagar/Patriot/Rasmussens of the world add value in the ensemble model--as they capture some subset of voters the others don't--but when the known to be biased pollsters comprise anything close to the bulk of the model, it opens the possibility of the system going screwy.
This is essentially my view. I've been "de-bouncing" the forecast by looking at Nate's electoral college bias chart, and mapping Kamala's current national poll lead to it. She's currently up by exactly 3, so the Electoral College bias charts would give her chances of winning at somewhere between 56% and 84%.
The RCP polling average is Harris at 1.8. Given what we know about Trump's overperformance in 2016 and 2020 vis a vis the electoral versus popular vote it absolutely seems credible to me that he could lose the popular vote this time out and still win the Presidency based on the EC.
Why use the RCP polling average instead of the very website you’re paying money to access specifically because it has better data?
Silver Bulletin has the national average at Harris +3, which is in line with the a very narrow Harris favorite/toss-up race.
I don't think it really makes a difference. What I focus on is what were the margins at this point of the race in 2016 and 2020 and what were the popular and EC margins versus polling in the actual election.
At this point in both 2016 and 2020 both Clinton and Biden had much larger leads than Clinton currently enjoys according to RCP. I would be surprised if Silver didn't show the same thing. The polling in both races narrowed at the end of the race. Again, I would be surprised if Silver didn't show the same thing. And the final vote margin for Clinton was +2 in the popular vote which meant that she still lost the EC by 56k vote. While Biden won by +4 nationally his EC margin was only 44k votes.
Is it plausible that Trump loses the popular votes by a couple of points this time out but still wins the EC? Absolutely.
Polling underestimated Trump in 2016 and over estimated Biden in 2020. It under estimated Dems in 2022 and dems in special elections in 2023/2024.
New voter registrations since July 21st in key demographics that are helpful to Harris shows that dems are being underestimated in the polls again and that Harris' win will be much bigger than people realize
Of course it's credible that Trump could win without the popular vote. I don't think that's what Grant was saying.
I'm just saying the current polling is consistent with that outcome.
Everyone seems to want to ignore this important detail
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4a
Real Clear Politics has Kamala averaging 2.6% which round up its 3%. RCP also excludes many polls and they are partisan. The polls are underestimating Kamala Harris and dems just like they did in 2022
What would have been a "typical" exposure-driven bounce got pulled forward this cycle...so pre-DNC forecast was unadjusted when it "should" have been adjusted, ergo the whipsaw from coin flip+ to coin flip-. The inevitable mean reversion always gets us home.
It's not going to be a coin flip. Kamala Harris is going to over perform the polls by a large margin
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
Another possible conclusion (which doesn't conflict with yours) is that polls are just more accurate this year than they have been in the past. They've figured out the cause of their misses in 2016 and 2020, and now they reflect the true state of the race.
What evidence is there that "polls are just more accurate this year"? YOu have none.
Nate has written about this extensively. Polls are challenging and while adjustments have been made, they may turn out to still be in error.
No, polls this year are going to underestimate Kamala Harris and the dems
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
This. The convention bounce was pre-empted by the transition bounce.
I agree with you the 'bounce' occurred after Harris entered the race. It was baked in by the time the convention came around. I also don't buy Michigan being Trump at this point in time, since few polls indicate that being the case. Once again, my own modeling work years in the past taught me to be careful with all these added 'adjustment' factors. They can get you into hot water fairly quickly.
They all are extremely biased. Patriot Polling has done like 7 polls total. And they got 2022 completely wrong
Compared to the big name polls that have a history of disastrous misses?
Yes Trafalgar, Rasmussen, and Patriot Polling have all had disastrous misses
Nate, you’ve mentioned a few times that betting markets are a sanity check on your model. Does it worry you that your model is starting to diverge from them? Polymarket has Kamala at 50% and predictit has her at 54%
https://www.realclearpolling.com/betting-odds/2024/president
lol betting odds. Allan Litchman has more credibility than Betting Markets
I'll take the wisdom of the crowds any day over a black box subjective 13 keys "model". Ask Nate what he thinks of Litchman.
lol wisdom of the crowds. Trump had bigger crowds in 2020 and still lost. But okay, don’t trust the dude who predicted every presidential election correctly since 1984. Go by crowd size lmao
So let's bet. Put your money where your mouth is. Name your wager.
I just did, are you slow and can't read? I said $1000.00
Betting odds of 60-40 one way vs 40-60 the other is not very different.
Point taken, but it's also about the trend. Harris was a five point favorite at one point. The trend is not good for her.
I don’t think she was ever a 5 point favorite (as in, a favorite to win the popular vote by 5 points). She was up to 55-45 or maybe even 60-40 in win probability. But as I said, that is very close to a 40-60 loss probability - only an odds ratio of 2.25 between those two probabilities, which is smaller than the difference between 60-40 and 80-20, or between 80-20 and 90-10.
Betting odds are a joke. Allan Litchman has more credibility than the betting markets
Polymarket is currently at Trump 51% Harris 47%. Expect that gap to widen after the debate.
Let’s be honest with ourselves here, it is not good for trump that a lot of lower information undecided voters are going to hear him formulate thoughts live for the first time in a while. I’m not a coconut pilled hater but it’s gotten bad lately.
They've been hearing him do that for almost a decade now.
I don’t want to veer into partisan trash talk but my view is:
1. The debate will draw many hundreds of thousands of people who primarily get their information from headlines/ clips and have not really tuned in to extended raw footage of either candidate since the last debate.
2. Trump, while not Biden levels of decompensated, is a much harder to follow than he was 8 or even 4 years ago. Less “what a jumbled way of getting to the point” and more “what on earth is he even trying to communicate.”
If it makes you feel better, I think the debate is unlikely to change the state of the race meaningfully. My original response is to the commenter who was projecting that it would, in Trump’s favor.
It really just depends on whether either one ends up looking completely foolish. It doesn’t have to be as bad as Biden’s performance, but if one of them seems comfortable and in charge while the other flails, it could make a difference. Ditto if either one cancels at the last minute. (I’m looking at you, Kamala.)
If they both manage to present a functional facade, then I agree it won’t make much difference.
In any case, it’s something we can all fight about with more data in the debate post-mortem. My forecast: we’ll all be declaring “it’s so over” for the one we don’t like.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
Yea Polymarket is dumb. Allan Litchman has better predictions than Polymarket
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
Since Nate took a stab at explaining the “why” behind the relative Trump rise, in association with RFK, and posited he and others may have played it down too much, I’ll ask a question.
Given the RFK voters you’ve met or encountered, how would you measure their engagement in this election?
If you answered anything other than “high”, you’re living on another planet. These people are tuned in. Which means they’re fully aware of RFK’s attempts to get off the ballets in swing states, and they’re keenly aware of how the Left (Democrats) are blocking him. From their perspective, he’s getting screwed again by the same people.
If the Dems had left it alone I suspect most RFK voters would have tuned out. Many would have stayed home on Election Day. Instead, what the Democrats have done is created yet another block of highly disaffected voters, and inadvertently turned them into Trump voters. Enthusiastic ones!
It didn’t need to be that way.
Yes, I'm sure these highly engaged voters would have stayed out of it but NOW they feel slighted and are motivated to vote.
This is pure unsubstantiated made up stuff.
It’s all over X. Tons of RFK supporters pledging to vote Trump. Dems reap what they sow.
That's funny. My Xitter feed is nothing but former Trump voters pledging for Harris.
It's almost as if social media curates content to match your own biases. Who knew?
Anything you say that begins with "Tons of RFK supporters" is already missing the point. There aren't tons of RFK voters, and if your feed is showing them to you, it tells you that your feed is not getting anything like a representative sample.
Twitter is a part of real life, but it's only a tiny fraction of real life. Anyone who posts there is highly unrepresentative.
"It's all over X" lmfao thanks for the laugh
agree
Ok, polls show RFK supporters go to Trump 2-1 vs Harris. Better?
Polls don't show that though
They do: "RFK Jr. supporters favor Trump over Harris by almost double (Trump 51%, Harris 26%), according to J.L. Partners/Daily Mail poll."
https://www.dailymail.co.uk/news/article-13773063/RFK-Jr-Donald-Trump-voters-kamala-harris-poll.html
At least for the few RFK voters I know, they are planning on voting for Harris now. A sizable chunk of RFK supporters were really just "double haters" that didn't want Trump or Biden.
Fwiw, my experience is the opposite. I have two friends who were enthusiastic RFK supporters. They are excited for an RFK lead health initiative under Trump.
There are very anti-Harris.
First, there is only a small minority of people who have ever considered voting for RFK. Out of those people, it's only a tiny fraction who are doing so because they are engaged. That tiny fraction of a small minority may be disproportionately active on Twitter, but they are tiny.
If the D's hadn't done petty crap like deny RFK Secret Service protection he may not have evolved into such a headache. It's bafflingly petty and stupid.
Secret Service protection is only provided to major candidates, lol.
Didn't RFK qualify? He should have qualified for matching funding.
No, matching funding is a very low threshold - only about $100k raised in at least 20 states which almost anyone can meet.
To qualify for secret service protection, a candidate must either poll above 15% for at least 30 consecutive days and be a member of one of the major political parties (a party that won at least 25% of the vote in the last election), or they must poll above 20% for at least 30 consecutive days if they are an independent candidate or minor political party (won between 5% and 25% in the last election). RFK was nowhere near this threshold.
So how is it that RFK ended up with secret service protection?
Biden himself directed the Secret Service to give RFK protection, for extra caution. But he definitely was not entitled to it. And he no longer has protection as it ended shortly after he suspended his campaign.
D's are fine. They are going to take back the House and Kamala is going to overperform the polls and win over 300+ electoral college votes
So... I only know one person that I know is an RFK voter, and that person is definitely not highly engaged in this election. They don't watch much news and they don't follow politics, but they have been dabbling in a very confusing swirl of new age philosophy, conspiracy theories, and off the deep end environmentalism. For example, this person believes plants are sentient. While going down internet rabbit holes they stumbled on some of RFKs work and found him to be the only politician they resonated with. I'm certain they are completely unaware of the nuances of DNC or RNC strategies.
Both you and Nate Silver are going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
That’s why you’re seeing the pop.
Nate and Eli, it’s time for your version of a snake graphic. I need to understand the paths to 270 Harris might have opened to go along with this “expected but bad news for Harris” post.
This.
"Electoral college bias looks like a challenge"
This seems to be a popular line of discussion on the left. From the outside it reads as mild derangement. Please let me know which part of the following you disagree with:
It is our election system. You're saying that our official election system is a challenge for Harris. An intelligent well-informed person saying that our election system is biased demonstrates an intentional disregard for the system by propagating the myth that the popular vote is what matters, when it doesn't. The popular vote is not our system by design. The endless discussions about the "electoral college bias" is reducing election validity in the mind of the audience. I grant you that it's a much milder form of what the right does, but it's really the same thing. You're saying that the election is not valid because it's biased against Harris. But that's not true!
That's not what "bias" means in the world of statistics, and by extension, how Nate is applying it here.
If we want to talk on a societal level, you're underestimating the cynicism-boosting effect of having an electoral system where the candidate the majority of the country wants to win loses in 2 out of the 3 most recent elections. What is there to encourage positive-spirited participation in a system where your vote has zero meaning? Why not just be out for yourself if most people can want one thing, and yet the most important decision is made by people in a handful of states?
Case in point. You're both claiming that the use of "bias" is statistical and following it up with an argument that the Electoral College should be replaced. Reality doesn't line up with preconceived notions all the time, but in this case it's essential to keep hammering on it.
I completely understand the argument for popular vote deciding the presidency. And yet I find the Electoral College to be a good idea. If you can't explain the rationale, then there's more to understand before tearing it down!
The place to start is with what goal you are trying to achieve. The founders' goal was "good governance", and they decided on a non-trivial election mechanism that was not even close to 1-person 1-vote. There are good reasons for republics instead of democracies, and much is written on the topic. Yet no one seems interested in that, just in saying over and over again how it's unfair.
The modern electoral college is nowhere close to what the founders intended though. Representatives to the electoral college were intended to act as a safeguard against an irrational public, as Hamilton articulated in Federalist Paper No. 68. The modern system of binding electors by state law emerged in the 19th century and is nowadays basically just raw mobocracy at the level of the state, instead of the nation. There are some potential reasons to value that, but the modern electoral college cannot be defended in reference to the founder's vision of republicanism.
Yes! That is an important change that was made. It sounds like we both agree that this change probably made the governance worse. My conclusion is that we should NOT take another step on that road (e.g. dispense with EC), but it sounds like yours is the opposite: that because we've already made that change, might as well make more?
I don't follow.
If you're asking for my personal view, it's that the modern electoral college system is worse than both pure Republicanism (which benefits from the decision-making of nominal experts) and pure Democracy (which benefits from being impossible to subvert to the whims of narrow interests). Instead, we have a system where the president is chosen by a narrow subset of laymen in a handful of random states, which encourages candidates to promise inane and wasteful patronage to these few individuals (see both Trump and Harris's proposal to eliminate taxes on tips in a bid to win voters in Nevada)
But the Electoral College isn't even remotely still functioning in the role it was intended to play. If we're already voting for presidents directly, which is what the Electoral College was supposed to *prevent*, what is its point?
I prefer it when people on the right simply admit that they want to preserve the Electoral College because it favors them electorally, rather than trying to drape themselves in nonsensical arguments about how it plays an important role in the system when it's already been completely stripped of its intended purpose. That, at least is a defensible position, as both sides are ultimately rooted in personal interest (though I think the Democratic position of "majority vote should win" is far more intuitive, & more in keeping with peer presidential democracies like France).
We get it, you hate democracy and think elites should be making the decisions
In your ideal "republic" you wouldn't be qualified to vote
Bias: it's a breath mint and a candy mint!
That's one potential disadvantage for the EC. But look back at history: the Articles of Confederation required a unanimous vote in its version of the national legislature because small states like DE did not want to be dominated by larger states like VA (the CA of its day).
Absent the compromise that the Senate and EC represent what incentive would small states have had for joining the Union? And doesn't that same logic hold true today?
No. States aren't going to secede bc the EC goes away.
From a purely objective viewpoint, the EC distorts the election by making some votes matter more than others. It's hysterical to me that the EC was ostensibly designed so that a small state's votes didn't matter...and what we've ended up doing is making it so that large state's votes don't matter.
The votes of large states obviously do matter, they've just been weighted to address the concerns of small states.
You may think it's unimaginable that states would secede if the EC were abolished, but by the same token isn't it just as unlikely that the country would ever move to abolish the EC? It's just not going to happen.
It's not that I "may think" it's unquestionably true. You think Utah's gonna give up all the benefits of being part of the US bc the EC changes. Never.
So that's a pointless statement.
Regarding the EC...there's already the National Popular Vote compact where states would agree to allocate their electoral votes proportionally based upon the popular vote in their state. This is gaining traction and over time could come to reality. Much, much, much more likely than any state seceding.
So much, much more likely than a 0.0001% probability?
At the end of the day you're asking smaller states to surrender political power in exchange for...what exactly? That's just not how the world works.
The EC was created not to balance small and large states, but to prevent voters from directly choosing Presidents, as the Framers feared that would be dangerous. It no longer functions in this role.
Moreover, the states which are currently benefiting the most from the EC in terms of political attention, such as Pennsylvania, Georgia, and Arizona, are not particularly small states.
Yes and the articles of confederation was a massive failuren
I take bias here as the term of art in statistical analysis rather than a value judgement.
Essentially, Nate means that there is a bias (or difference) between the popular vote and the electoral college result.
If you look at the thread above your message, you can see that other people do not take it in a statistical manner but instead are getting riled up about how the Electoral College is unfair. It demonstrates my point perfectly.
It is unfair. That's indisputable. Any system that makes one person's vote more valuable than anothers is inherently and unquestionably unfair.
You're mistaking their objections to the EC as a statement about "bias in the polling". They're two different things.
The goal is "good governance", not your limited notion of what's fair.
Is it "fair" that a bum on the street have an equal vote to Kamala Harris herself? Why is that fair? The bum may not even be able to read, know nothing about the economy, know nothing in general. He may be days away from a Fentanyl overdose. Why is it fair that he gets to have a say about the rules that I have a live under?
Why would it be more fair for the populous city centers to unilaterally set rules that rural workmen on the oilfield in the middle of the country have to live under? The votes representing their interest wouldn't matter in that case.
The 2 Electoral College victories in the past 2 decades with a minority of the popular vote gave us the Iraq War and January 6th, and this time might restore a president who attempted to overturn an election with mob violence against the Congress, which was one of the Framers' worst nightmares. That doesn't look like "good governance" to me.
The system is completely fair. The rules are set out ahead of time, and (relatively) impartially adjudicated by the courts.
The electoral college has a purpose, as anyone familiar with history knows. That purpose is to require a presidential candidate to both get a large number of votes and to have a majority in a large number of states.
Regardless of how I feel about the election result, it has served its purpose in recent years.
Don’t mistake undemocratic for unfair. Many things that are less democratic (e.g. an independent judiciary) serve a valuable purpose (said judiciary protecting rights from policies supported by the majority of people).
The electoral college doesn’t require a majority in any state. Look at a three way race (for example, 1992) to see clearly why the electoral college doesn’t do what you claim. The winning candidate only got a majority of the votes in two states that year: Arkansas and DC. The winning candidate received less than 40% in seven states.
*Edit for clarity: Winning candidate above refers to winner of a state’s electoral votes.
Just because everyone knows the rules ahead of time doesn’t mean it is a fair or reasonable system. If we wrote new rules that said coastal states got double the weight of interior states because of how important their port connections are, that could be a fair system by your logic but it would be a terrible system for democracy.
The purpose of the EC, as intended by the Framers, was to prevent voters from directly selecting a President, fearing that that would have a corrupting effect on the country's politics. When I go to vote, I mark a vote directly for a President, not electors. It no longer functions in its intended role.
Nate can defend himself and his verbiage without my commentary... BUT the "bias" is in reference to the popular vote margin not her personally. The Electoral College is completely valid - but in terms of speaking statistically it's fine to use the term "bias" when discussing the correlation between national polling vs EC votes. As a nation we generally speak in terms of national polling ex. "He's winning by 5 points"- so when that does not coincide directly with the EC margin - it is 1) interesting and 2) a strategic campaign topic to discuss
> As a nation we generally speak in terms of national polling
Yes, a focus on the popular vote when that is not the election mechanism and known to be misleading in predictable ways is a choice that someone made. I certainly agree with that. I think that reinforces my point though.
(I agree that it's unfortunate that there isn't a single number that can be easily reported for the balance of Electoral College. It would be much stronger evidence of intentionality if there were and the MSM simply chose not to use it)
> in terms of speaking statistically it's fine to use the term "bias" when discussing the correlation between national polling vs EC votes
So why isn't it the popular vote which is spoken of as biased rather than the EC?
Your first point answers your second
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
There are two elements of how the EC works and which give voters in smaller states disproportionally more power that have nothing to do with its creation in that they weren’t how it always worked and were not the intention. The first is the winner take all proposition. There is nothing in the constitution that says each state should award its electors that way (and a few don’t). Additionally, the framers never contemplated the House (and therefore the EC) having a static size. If the House had continued to grow as it did for the first part of American history, in line with population, the small state bias would be dramatically smaller.
Interesting point about the limited size of the House being an important factor. I hadn’t thought of that. The size-related skew due to the Senate would be much smaller now if the House had grown to have several times as many seats.
I also say that the official election system of Iran is a challenge for reformist candidates. An intelligent well-informed person saying that the Iranian election system is biased demonstrates an intentional disregard for the system by propagating the myth that the preference of the people is what matters, when it doesn't. All that matters, according to the Iranian system, is the vote of the people among the choices approved by the Ayatollah. That is by design. The endless discussions about the bias of the Ayatollahs is pointless moralizing. We shouldn't care what is right when doing politics - all that matters is how the systems are set up.
One doesn't have to think that the Iranian elections are "not valid" to raise objections to how they are carried out, and the same is true for the United States. I don't recommend a revolution in either country - I recommend reform (even though the system is set up in a way that makes particular reforms very difficult to carry out).
Great analogy to the EC arguments being put out here. I think you also hit on the area that probably causes some of the pushback: inability to distinguish between a system that should be improved to be more democratically fair versus cheating or having a “rigged” system. Trump becoming President with fewer votes due the EC disproportionately benefitting him doesn’t mean he cheated.
RE "Why does the Electoral College work this way":
> America’s complex voting system was set up to provide shares of political power, not to simple head counts, but to an intricate network of society’s functions and interests.
> Election of senators by the legislative bodies of the states was intended to tie their representative interests to the states, rather than to national political parties.
> There are many more examples, but the principle is clear: the Constitution is not a poor attempt at equal representation of people, but a complex compromise dividing political power among interests, all of which are required for a functional republic, and which have political needs and interests which they require political power to defend from legal interference.
> So, yes, the "flyover states" are overrepresented if you want America to be a pure 100% democracy, which it never was, and was never intended to be.
> But they are also the reason why you have nice things.
https://substack.com/@devoneriksen/p-147280453
Imagine if the US had never been formed because smaller states in 1787 had refused to join a country where larger states would have dominated the national dialogue.
If you go back and read your Federalist Papers, the EC wasn't just about small state v big state representation. It was also based on the idea that regular people are too stupid to vote for president directly, so they need to elect elite electors instead, who can select a president without being driven by inflamed passions and demagoguery.
We have rightly discarded that as antidemocratic nonsense and now electors are expected to vote for whoever the citizens selected on their ballots, but the EC persists as an outdated relic of 18th-century views.
It's not an either-or proposition. Even if the EC is an anti-populist measure why do small states get a disproportionately larger share of electors? That has everything to do with the small state to large state dichotomy.
And that argument also ignores the issues that would pop up today if the EC and Senate were disposed of. What incentive would smaller states like Utah or Alaska have to participate in that kind of federal union?
If that were really an issue today, I don't think Vermont, Maine, Delaware, Rhode Island, and DC (edit - and Hawaii) would all be part of the National Popular Vote Interstate Compact. Membership in NPVIC correlates strongly with partisanship of a state, but really not at all with size of the state (except insofar as size of the state correlates with partisanship).
Let me know when AK, UT, WV, etc. sign on.
You say that as if we'd miss them :)
On a more serious note, you're correct that small states would never agree to change the system, and that's why it persists. But that's not really relevant to the question of whether the system is fair and democratic. Why is my voice more valuable if I move to Vermont than it is if I move to New York?
Because if that original compromise hadn't been made you would probably need 50 votes in a 50 seat national legislature to pass anything.
Wow. Are you really so obtuse to not understand why Utah or Alaska would want to be part of the US even without an EC?
Because the number of reasons is incalcuable and far, far outweigh whatever negatives you think are associated with abolishing the EC.
You're either extremely narrow-minded in these areas or, as I suspect, not addressing this honestly.
You guessed the answer on your second try.
Unfortunately, this is a pointless exercise in attempting to have a good-faith discussion.
It depends on what membership in a country without EC style safeguards would entail. What if CA/NY moved to significantly curtail petroleum drilling? Would AK be okay with that? I doubt it.
200+ years ago, it really was about states wielding political power, and people saw themselves more as citizens of their states than of the US. The system was created to balance Delaware's political power against Virginia's, not to give a citizen of Delaware more votes than a citizen of Virgin. Neither of them really got one anyway.
Go hang out in some oilfield in Texas, the UES in Manhattan, Venice Beach in CA, Provo in UT.
The US is a staggeringly diverse place. In my mind a system that allows for significant state representation is a good thing given that diversity in that it means that only policies that have significant general support ever make it into law.
The first chart showing the change pre DNC to now is calculating the change wrong on several of the states.
I'm not seeing that. Where in the cart is it wrong?
The static image posted is incorrect, see North Carolina going from D+0.6 to R+0.8 but the change is shown as R+1.6, for example. It looks like it's fixed if you click through to the "interactive chart" but it stoll presents wring in the sustack app and in the emailed version.
Ah, that's why I'm not seeing it. I'm on desktop, which is presenting the chart in html format.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
I just looked at Lichtman's model at wikipedia. https://en.wikipedia.org/wiki/The_Keys_to_the_White_House
How anyone could call "Trump" an "Uncharismatic Challenger" is beyond me. That makes 5 falses, not four, and very close to the tipping point... especially when people don't necessarily view the short or long term economies as strong, nor do they likely recognize any recent military success.
The dude who lost the popular vote twice and averaged 41% approval for his entire presidency is charismatic? How about you watch the video that gives his reasoning for each one. Charismatic candidates are those with cross over appeal that is well beloved. Trump rambles about Hannibal Lector and is hated by 60% of the country. Also Allan is using the same metrics he has used since 1981. He's not going to change it just because some random person on substack thinks Trump is charismatic. No, its four.
Watch the video, he explains it. Short term and long term economy isn't what peoples public perception is, he's looking just at the raw data. Are we in a rececession? No. Again he's using the same metric that he's used since 1981 to make the predictions. Your opinion on it doesn't matter. Cope
https://www.youtube.com/watch?v=JoWt1EOA340
Wrong!!!
Not wrong at all. Cope
Wrong, Nate's model in 2016 was similar to the other forecast models around the election day. In the summer of 2016, Nate also had a 90-95% chance of Hillary winning like all the other models. Its a myth to claim Nate Silver's forecast was somehow better than the other forecasts. Like I said, I'll take Allan Litchman's predictions since he has gotten every presidential election correct since 1984.
https://www.youtube.com/watch?v=xE22XjWEyQE
As for judging people by how they vote and the color of their skin. I did no such thing, TargetSmart increase in voter registrations came after July 21st which is when Harris became the nominee and Biden dropped out. Kamala has always done better with black people, hispanics, and women, then Biden. It's obvious that when she became the presumptive nominee on July 21st that the surge in new voter registrations are people that would more likely vote for Kamala. Its not judging anything. Its just a fact, black people and hispanics, especially women, vote for democrats by much larger margins than they do republicans
Lol typical name calling. If what you were saying is true you would provide the evidence but you don't have crap. Cope and seethe. Trumps going to lose. No criminal in the WH for you
Nate, with pro-Trump and anti-Trump presidential juice sucking up all the political forecasting oxygen, there has been next to nothing, here or elsewhere, on Senate and House forecasts. There really needs to be a lot more focus and coverage of that.
For instance, whether or not Harris is able to get over the 270 electoral college hump against Trump, the left's enthusiasm for her over Biden has to have shifted the landscape somewhat on Senate, and particularly House forecasts, including with respect to which party will control each of those chambers, and I would like to see content that analyzes that. Regardless of who is elected President, control of those chambers is going to have a monumental effect on the direction of the country -- e.g., a Trump presidency would be vastly different under Republican control of the Senate and House than Democratic control.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
Cook updated its House race ratings today. Some of it is paywalled, but some interesting analysis there and more interesting charts on Dave Wasserman’s twitter.
Not to cut down on subs, but it's clear a lot of people would be happier if they just wrote "Kamala Harris will win" on a piece of paper, then looked at it every day.
The polls are the polls, they've been wildly off for two Presidential elections straight, particularly for the entire campaign before October (great that they were only off by a point on election day 2016, but the seven months leading up to that had voters being told she was leading by 6-10 points) and for state polls in general (claiming Clinton was up 2 in Ohio when she lost by 8).
The polling data is far better used as a topping for the fundamental data we have about right track/wrong track, voter issues, and the economy. It's idiotic to pretend as if it's likely to be accurate, it's much more likely to not be, and the people with the greatest political literacy all know this.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
Highly unusual circumstances under which Harris gained the nom. Wonder if her convention bounce was after Biden stepped down and now she is just regressing to the mean.
I was really surprised when I saw how dismissive Nate was of the effect of RFK Jr dropping out - both he and the national media seemed to despise the guy to a degree they willfully assumed a candidate with 5% of the vote dropping out wouldn't move the needle. I nearly left a comment on Nate's original coverage of RFK arguing that it was going to have a more significant impact than said, and now I wish I had because I would have been proven right, lol
not just dropping out, but endorsing Trump I think could easily have a 1-2 point move
But it didn't.
RFK supporters break 2-1 for Trump. Will absolutely have an impact in close swing states.
yeah. I suspect that what happened is 1)Nate + national media weren't expecting this to be a dead heat race, and in previous races that *weren't* close, 3rd party candidates dropping out hasn't historically had a huge impact. So the first mistake was in assuming this wouldn't be a close race (I think they keep having this knee-jerk reaction of expecting elections to go back to pre 2016 patterns) and 2)They assumed that RFK's poorly-run campaign somehow translated to "no-one cares about him", in the face of that literal 5% staring them in the face. You can despise a candidate, but to let that affect your assessment of impact is a very silly thing to do
Its not a close race. Kamala is going to win by a large margin. Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
Do you think that's likely? I would expect more like 2 for not voting, 2 for Trump, 1 for Harris. But RFK was lower in the swing states to begin with than nationally, so this matters, but by less than people think.
Sure, and I may be reading it wrong, but the thrust of this post seems to be “the reason Harris is behind in my forecast is because of bad swing state polls, not because of the convention bounce.” Which isn’t true. If it weren’t for the convention bounce adjustment, she’d be ahead, so people are right to point out that the reason she’s behind in Nate’s forecast is the convention bounce adjustment.
Try less than a percentage point if anything
They do: "RFK Jr. supporters favor Trump over Harris by almost double (Trump 51%, Harris 26%), according to J.L. Partners/Daily Mail poll."
https://www.dailymail.co.uk/news/article-13773063/RFK-Jr-Donald-Trump-voters-kamala-harris-poll.html
I suppose that’s 1 for not voting, 2 for Trump, 1 for Harris, rather than the 2-2-1 split I was imagining. But I also am not at all convinced that a poll like this accurately measures how many will fail to vote.
Nope, supporters don't break 2-1 for Trump. He dropped out over two weeks ago. Polls have only been tightening within the last week
They do: "RFK Jr. supporters favor Trump over Harris by almost double (Trump 51%, Harris 26%), according to J.L. Partners/Daily Mail poll."
https://www.dailymail.co.uk/news/article-13773063/RFK-Jr-Donald-Trump-voters-kamala-harris-poll.html
lol you posted an article from August the day he dropped out or before and you think that is evidence? And its a DailyMail poll which is not reliable, DailyMail is a right wing tabloid. You can clearly see that RFK didn’t help Trump at all.
Trump lost the election in 2020 by 44k votes in three states. Assume the same general margins this year. What if RFK adds 10k voters to Trump's total in those three states? In my mind that's an extremely modest proposition for somebody who's polling at about 5% nationally.
100%. It was just common sense to assume that RFK WOULD move the needle. there's a difference between him dropping out not having a huge/sizeable impact - no-one expected that a candidate with 5% vote would drive a 10% swing or any - versus having not even a 1% impact, which the media projected for no reason other than...wishful thinking on their part I assume
For a 5% candidate dropping out to have a 1% impact, you need a very strong effect. A substantial fraction of those people weren't actually going to vote to begin with (that is always the case with people who poll as supporting a third party candidate) and once he's dropped out they are even less likely to vote. You need a very lopsided break for a fraction of 5% to make a 1% difference on net.
5% of a 100,000,000 voter turnout comes out to about 5 million. 30k votes out of that pool would seem to be a pretty modest proposition.
That's 5 million total across all 50 states. About 20% of voters are in the swing states, so that's about 1 million total across all 7 states, or about 200,000 in any one big swing state (assuming that swing state voters are just as likely to vote third party as voters elsewhere, which seems like a bad assumption).
Getting a margin of 30,000 out of 200,000 requires something close to a 60-40 split, with no abstentions. It's not out of the realm of possibility, but not particularly likely.
60-40 is within the realm of possibility just in terms of the "natural" voter split for RFK supporters. Now add in an endorsement? Definitely plausible
You’re talking about AZ (~11k), GA (~11k), & WI (~20k). It’s plausible for Trump to flip all three with RFK Jr endorsement but Harris is polling pretty well (+3.2 avg per 270toWin) in WI, and Kennedy in WI seemed more like 4% avg. before he dropped out.
How were Biden and Clinton polling in those states at this point in time in 2020 and 2016?
Good point. Man oh man the 2020 WI polls underestimated Trump so badly. Like by ~7%. On the other hand, 2022 WI Senate polls were pretty accurate, although they did underestimate the very weak defunder D candidate Barnes, who ended up losing by only 27k even though polling average was -3.4%.
I would say the 2022 midterms were in general more accurate than either the 2016 or 2020 presidential polls, but they are different animals.
WI historically is difficult to poll, and has polled too D in previous cycles.
Trump lost PA by 80,550 votes. Lost Wisconsin by 20,682 votes, and lost Michigan by 154,188 votes. Even if RFK adds 10,000 voters for Trump in each state. Trump would still lose
RFK dropping out didn't have any real impact. Kamala Harris was still seeing increase leads even after RFK dropped out. Its only been tightening within this last week.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
On June 26, your Biden vs Trump model had Biden's odds of victory at 33.7%. You wrote, "As our model launches, either Biden or Trump could easily win — but the odds are in the ex-president’s favor." The headline was "the presidential election isn't a toss-up."
Today's Biden vs. Harris model has Harris' odds of victory at 38.3%. Given the similar odds of victory for Biden on June 26 as Harris on September 6, could you write the same headline you wrote about Biden about today's race?
"Either Harris or Trump could easily win — but the odds are in the ex-president’s favor. The presidential election isn't a toss-up."
I have been thinking the same thing, but I think Nate is waiting for more state polling to confirm [PA]. I'm wondering if 2 or 3 decent Kamala polls by the right pollsters would wildly swing the forecast.
My sense is that the post-convention bounce adjustment is a little overly aggressive, but I’m not too exercised by it.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
I think Nate is constantly thinking about how the model works and what information it doesn't have. He has been clear that he doesn't like to make adjustments to the model mid-stream, but that hasn't stopped him from highlighting when he think the model may be temporarily biased in one direction or the other. For example, after the debate Nate was clear that he thought Biden's position in the model was overly optimistic.
You're talking about June 26, which was one day before the debate, but even at that point I think Nate had a sour enough view of Biden that he might have been applying some mental downward pressure to Biden's chances. I mean, Nate was saying Biden should drop out long before most people.
Also, 33.7% is not the same as 38.3%. You have to draw the line from "toss-up" to "not-a-toss-up" somewhere, and where you draw that line may very well be between those two numbers. 35% for example. I mean, maybe it's 40% and the race has moved into non-toss-up category, but if I were Nate I'd wait Harris has been under 40% for a week before I declared "not a toss up"
I think the choice to apply a penalty to adjust for a postconvention bounce is a defensible one, I just think it’s open question how applicable it is to this particular election. We’ll find out soon.
Yeah I think that's a very fair point, and is probably another reason Nate isn't declaring it "not a tossup." He's stated many times that if Kamala's current polling lead holds her win% in the model will get better over time.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
We’ll find out soon enough!
Yep and we will find out that I’m correct and Kamala will win lol
It’s not stubbornness. It’s process. If you adjust the model midstream as you see different things happen but before getting any actual results, you aren’t modeling based on statistics, you’re being a pundit. Nate sets his own rules at the start of the process, that unless there is something that is clearly and obviously wrong/broken, he doesn’t adjust the model midstream, he makes adjustments between elections. In this case it is highly possible that this is leading the model to be off at the moment ( although if so it will wash out long before the election) but saying he isn’t making an adjustment because of stubbornness is just a really strange perspective.
Nope. Its stubbornness. Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
I guess you're not taking into account what we saw in the Kansas reference after Dobbs - that enthusiasm among women - of all ages, but especially 18-29 - is very high and they may come out in numbers larger than any poll is predicting. Turnout is very hard to predict. Actually I wouldn't call it "enthusiasm". I would call it female "RAGE". Donald Trump offers us nothing. And we have everythihng to lose with Trump and the MAGA GOP. So there's that. Put that in your model and smoke it.
And how's your sugar-daddy Peter Thiel doing? Got anything bet on the race yourself, Nate?
Take that partisan vitriol somewhere else, this is a place for productive debate and discussion.
Their tone aside, you’re an idiot (or worse) for being so dismissive of their point. 😔
Your ad hominem argument is very persuasive.
Nate Silver is going to be so wrong!
Check the mirror
🤣🤣
looking fine over here. You should let go of that anger, have a nice life!
Kamala Harris is going to win big
🤣🤣🤣
You’re admiring all those well reasoned out arguments in your OP?
You blazing hypocrite
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
😔
Loss of privacy of your body isn’t a huge deal? I expect you haven’t actually put thought into this
Obviously you didn’t use the word because you haven’t a clue what it means?
That last phrase also makes clear it isn’t the only thing you’re mentally challenged on. 🙄 Wound up by made-up BS
Your current state polling averages have Harris narrowly ahead in PA, WI, MI, GA, and NV. But your forecasts have Trump narrowly more likely to win all of these states, and therefore more likely to win the Electoral College. The inescapable conclusion is that your model assumes we are about to see a shift towards Trump in PA, WI, MI, GA, and NV (as well as nationally). Right?
It's unclear to me how much of that assumption in the model — that things are about to shift towards Trump — is due to the convention bounce adjustment. Are there reasons other than the convention bounce adjustment that the model assumes things are about to shift towards Trump?
If not, if that assumption is 100% due to the convention bounce adjustment, then am I correct in thinking that the model would have Harris favored to win if all the polling averages are exactly the same in 2 weeks as they are today?
I believe yes, that's what Nate says about the adjustment towards Trump "it will begin to work its way out of the model over the course of the next week or two"
Imagine, hypothetically, that 2 weeks from today, state and national polling averages have moved towards Harris by just 1 percentage point relative to where they are today. In this hypothetical scenario, her win probability in your model will have spiked — from under 40% today to well above 50% in 2 weeks. Right?
Won't people be confused as to why that is happening, if you aren't clear now that her current win probability is being lowered by the convention bounce assumption?
From your point of view, as someone who has to explain what the model is doing to everyday readers, it seems to me like it's in your strategic interest to be clear and open about the effect of the convention bounce assumption right now!
It seems to me that if you ran a version of your model today with NO convention bounce assumption, it would show Harris as the narrow favorite to win. Am I right in thinking that? Why would it be a problem for you to admit that that is the case, while still saying that you think the convention bounce assumption is the right assumption?
That's what this post is for - to make clear that the model is building in a convention bounce adjustment.
Sure. I guess I am responding partly to Nate's saltiness on Twitter, where he has been very annoyed by people correctly pointing out that the only reason his model has Harris behind right now is this convention bounce assumption.
I think it would be helpful if he clearly stated the obvious, which is that his model would have Harris as the favorite to win right now if it weren't for that one specific assumption – which is in his view good and defensible assumption. And which, in his defense, is an assumption he has built into the model every 4 years for as far back as I can remember, so it's not like it's something he's added in capriciously or for any nefarious reasons.
Instead, he seems very determined to point out problems for Harris that are NOT the convention bounce assumption. I’m not denying those problems are real – of course it’s bad that her leads have slightly declined in state polling. But when your model is a bit of a bullish-for-Trump outlier among models, as Nate’s is right now, it seems disingenuous to focus on reasons for that bullishness that are anything other than “the model assumes Harris is in the middle of a convention bounce that is about to fade.” Because if the model didn’t assume that 1 specific thing, it wouldn’t have Trump favored to win.
He's stated in multiple articles here that Trump's current lead in the forecast is due to the model applying a DNC convention boost. I haven't read his X posts about it, but from what I've seen here, he has been very clear about that.
I remember an article here recently saying that if the polling today were the same the day before the election, the model would have Harris as a clear favorite. I'm not sure how much clearer he can be.
Yeah, it’s true he has been better about this here than on Twitter. For context I’m talking about tweets like these
https://x.com/natesilver538/status/1832060619488698429
https://x.com/natesilver538/status/1832098218710024700
And in fairness, I know Nate gets a lot of crazy unhinged comments on Twitter and I’m sure it’s hard not to let that get to you.
But regardless, as Nate says in this post, it’ll all be moot in a couple weeks when the convention bounce adjustment is mostly gone. So oh well!
He has a new post showing what the numbers would look like without the convention bounce adjustment. It's an interesting read.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
Yep, that's what the convention bounce part of the model does
My question is are all the people, especially boomers who died since 2020 and the surge in recent registrations among Dem leaning voters factored into the PA and NC numbers? I admit that I am severely uneducated in statistics and modeling.
As I understand it, the forecast assumes that things like changes in the electorate and enthusiasm levels will ultimately be reflected in polls. Lots of pollsters already weight responses to account for demographic shifts, so you wouldn’t want a model to double down and over-skew as a forecaster.
Nate Silver is going to be completely wrong. He's the only forecaster predicting Kamala losing, 538, Economist, Race to the WH, etc all have their forecasts showing Kamala winning. Allan Litchman who has predicted the winner of every presidential election since 1984 says Kamala is going to win.
Nate got 2016 completely wrong and also got 2022 wrong as well. This is just sad
Kamala is going to win much bigger in 2024 than people think. Polls under estimated Democrats in 2022 and in the special elections in 2023 and 2024 which showed democrats drastically over performed what the polls were saying. Since July 21st, there's been a massive surge in new voter registrations in key voter groups that are favorable to Kamala
"According to Target Smart, data show that registration is up 175.8% among young Black women; 149.7% among young Latinas; 98.4% among Black women overall; 85.8% among Black Americans generally; and 83.7% among young women as a whole."
https://www.salon.com/2024/08/29/with-democrats-fired-up-by-harris-candidacy-young-people-are-now-registering-to-vote-in-droves/?fbclid=IwY2xjawFI5HRleHRuA2FlbQIxMAABHWOHktQtYrEE5bAWhL2lKtLcBrnJh5LZWNTA6vvZdKe05-zaB3-0V9wgyw_aem_dkT9JWiK-erjaGbuP-O4aA
A like for quality analysis and clear exposition -- still a depressing read.