Yes, you're smarter than both Nate Silver and Malcolm Gladwell. Take the rest of the day off!
Malcom Gladwell doesn't understand why Jeffrey Toobin was fired (spoiler: he was caught playing with his little Toobin on camera in a work meeting), thinks Joe Paterno did nothing wrong, and is a Salazar apologist. "Being smarter than Gladwell" is a low bar to clear.
Rojo, there is 100% probability that Nate Silver is smarter than you
Close to 100% that he's better at stats than me although my lone A+ at Princeton was in stats.
But overall intelligence? I'd put it way less than 100%. Does he believe a man can give birth?
PS. The whole reason I started this thread was even if he thinks the Senate is 98% correlated with the House, it would drop his percentage by 1%. He's not dropping it at all. So I guess in his mind it's at least 99% correlated.
The numbers you're citing are "joint probabilities", not "correlation". The terms are related but not interchangeable.
At the bottom of 538 Election Forecast page, one can download model output files and look at the joint probabilities of outcomes, so there's no need to guess.
As of 18:05:08 27 Oct 2022 the projected probability of the Republican Senate & Democratic House outcome was a bit under 0.5% per the 538 "Deluxe" Model and lower in the other two.
PS. The whole reason I started this thread was even if he thinks the Senate is 98% correlated with the House, it would drop his percentage by 1%. He's not dropping it at all. So I guess in his mind it's at least 99% correlated.
You still don’t understand the numbers you’re critiquing.
PS. The whole reason I started this thread was even if he thinks the Senate is 98% correlated with the House, it would drop his percentage by 1%. He's not dropping it at all. So I guess in his mind it's at least 99% correlated.
You still don’t understand the numbers you’re critiquing.
He never will, either. I know for a fact he did not get an A+ in ANY class at Princeton
Nate Silver of fivethirtyeight became famous as he once predicted all 50 states in a presidential election right. ESPN gave him a ton of money but it now seems that the site is never right.
I was looking at it this morning.
He says there is a 46% chance the GOP takes the Senate. He says there is an 82% chance they win the House.
So what are the odds that the GOP also takes the House and the Senate?
He says it's also 46%. WHAT?
That makes no sense. I know the two aren't totally independent events but still that's crazy. It would have to be less than 46%.
The way I read that is he thinks there's a 100% chance the GOP takes the hosue but won't admit it.
rojo wrote: The whole reason I started this thread was even if he thinks the Senate is 98% correlated with the House, it would drop his percentage by 1%. He's not dropping it at all. So I guess in his mind it's at least 99% correlated.
Let's say you pick 10 D1 high jumpers, and have them make 2 attempts: once at 1.85m, and once at 2.15m.
After looking at previous results, you predict that at 5 jumpers will clear both heights, 3 will clear only once, and 2 will not clear either. Based on this, it is clear that the outcomes of an athlete's 2 jumps are not "99% correlated". After all, 30% of the jumpers are predicted to have different outcomes on their 2 jumps.
Now, what is the chance that someone clears the 2.15 but doesn't clear the 1.85? Not impossible, but pretty low! Less than 1% I'd say. It is much, much more likely that any "1-for-2" guys end up clearing 1.85, but not 2.15.
That is the situation here. According to the models 538 created, the senate is the high bar, and the house is the low one. GOP might end up in control of 0, 1, or 2 chambers. There is a real chance of each of those...according to the models. But, in the case that Americans elect enough GOPers to give them control of the senate, those same Americans are overwhelmingly likely to also turn the house red while they are at it.
Of course the models, and the inputs into them, might be good, or they might be hot garbage. Don't know, don't care. Either way, there is no "basic aspect of probability" that has been violated.
A simple "Who's ahead?" (even if by less than 1%) in the polling averages has the Senate at 50/50. But you have to consider that
* Polls have been underestimating Republican support by a larger amount each election season. 2020 was almost a joke; the polls said Biden was up double-digit %s on average in Wisconsin and he won by like 0.2%.
* The momentum is in the Republican direction. You see races where the Republican candidate was trailing and recently caught up in the polls, but you don't see races where the Republican was leading and then falls behind.
It's highly, highly likely that the Republicans pick up a seat in the Senate. Nate Silver still says it's less than likely.
Polls typically underestimate Republican support in Pennsylvania by about 6%, for example. If Fetterman were ahead on average by around 4%, that wouldn't be anything close to a comfortable lead, would practically be a wash.
rojo wrote: The whole reason I started this thread was even if he thinks the Senate is 98% correlated with the House, it would drop his percentage by 1%. He's not dropping it at all. So I guess in his mind it's at least 99% correlated.
Let's say you pick 10 D1 high jumpers, and have them make 2 attempts: once at 1.85m, and once at 2.15m.
Well with jumping there is a chance a guy has a bad jump in a one off. This example is more like
NAU has a 46% chance to win nationals.
NAU has an 86% chance to have at least 3 AA.
What are there chances of winning nats and having 3 all Americans ? RoJo wants to combine them and say the odds are like 40%. But in the real world we know if NAU doesn't have 3 AA their odds of winning nats is close to 0%.
The amount of things that would have to break the repubs way to win the Senate but not the house is enough to be a rounding error.
Nate Silver is nothing but a low-rent mental health counselor for liberals. His job is to make them feel slightly less bad about obvious facts. The downside is that libs are even more devastated when reality arrives. Nate has been comically wrong these last two months about what will happen on Nov 8.
Historically 538 forecasts have been pretty well calibrated. That means when they say something will happen X% of the time, it happens about X% of the time. This is true in both sports and politics.
Here is a page with some data on this, admittedly I'd prefer something from a third party but the methodology they describe sounds reasonable:
Yes, you're smarter than both Nate Silver and Malcolm Gladwell. Take the rest of the day off!
Malcom Gladwell doesn't understand why Jeffrey Toobin was fired (spoiler: he was caught playing with his little Toobin on camera in a work meeting), thinks Joe Paterno did nothing wrong, and is a Salazar apologist. "Being smarter than Gladwell" is a low bar to clear.
