Donald Trump

Vantage Data House

app.vantagedatahouse.com app.vantagedatahouse.com

This makes for a very interesting read, and it highlights something that even I have overlooked despite my slightly obsessive tea leaf reading.

The poll aggregators (538/RCP) are currently predicting something which seems functionally impossible. Split ticket races in the majority of swing states. Split ticket races are exceedingly rare - and becoming rarer with time due to tribalism. In fact over the last two elections there has only been one single instance where a Senate seat went one way and the Presidency went the other. Simultaneous Senatorial races and Presidency races historically correlate to 95% accuracy in terms of outcomes. So about a 5% chance of a split ticket happening.

Yet… both of the main poll aggregators are showing this happening this year in Michigan, Nevada, Arizona and Pennsylvania. All of the Republican senators are polling way worse than Trump. Even Cruz is sweating about his seat. This really does make zero sense. Based on historical averages there’s about a 0.6% chance of that actually happening.

Some might be thinking - “well Trump is a bit of an outlier and tends to outperform GOP Senators”. But this isn’t accurate, in the 4 swing states that had senatorial races in 2020, he outperformed the Senator in 2 and underperformed in 2 and the average difference between the GOP Senator’s vote share and Trump’s vote share was less than 1%.

So with this in mind how is it possible that 538/RCP are predicting 5-7 point losses for GOP Senators but wins for Trump?

The answer is that the poll populations are different. And the methodologies behind those polls are different. And what we’re seeing is the prediction of two very different sets of results. One where it’s not even close and Harris wins comfortably and another where Trump wins by a hair.

It also highlights the huge swing in the gender gap which has already been mentioned. Women are turning out both in polling and in early voting.
 
SkyBlueFlux said:

Vantage Data House

app.vantagedatahouse.com app.vantagedatahouse.com

This makes for a very interesting read, and it highlights something that even I have overlooked despite my slightly obsessive tea leaf reading.

The poll aggregators (538/RCP) are currently predicting something which seems functionally impossible. Split ticket races in the majority of swing states. Split ticket races are exceedingly rare - and becoming rarer with time due to tribalism. In fact over the last two elections there has only been one single instance where a Senate seat went one way and the Presidency went the other. Simultaneous Senatorial races and Presidency races historically correlate to 95% accuracy in terms of outcomes. So about a 5% chance of a split ticket happening.

Yet… both of the main poll aggregators are showing this happening this year in Michigan, Nevada, Arizona and Pennsylvania. All of the Republican senators are polling way worse than Trump. Even Cruz is sweating about his seat. This really does make zero sense. Based on historical averages there’s about a 0.6% chance of that actually happening.

Some might be thinking - “well Trump is a bit of an outlier and tends to outperform GOP Senators”. But this isn’t accurate, in the 4 swing states that had senatorial races in 2020, he outperformed the Senator in 2 and underperformed in 2 and the average difference between the GOP Senator’s vote share and Trump’s vote share was less than 1%.

So with this in mind how is it possible that 538/RCP are predicting 5-7 point losses for GOP Senators but wins for Trump?

The answer is that the poll populations are different. And the methodologies behind those polls are different. And what we’re seeing is the prediction of two very different sets of results. One where it’s not even close and Harris wins comfortably and another where Trump wins by a hair.

It also highlights the huge swing in the gender gap which has already been mentioned. Women are turning out both in polling and in early voting.
Click to expand...
Great find though you’re right, they won’t be split.
 
BlueMoonAcrossThePond said:
And with respect...
You (and I too) are very anti-Trump.
The tendency is to view evidence in support of ones convictions in a favorable light; and those in opposition to ones views, as suspect.
The truth is - there are numerous credible arguments about why the polls are off, inflating the support for Trump; and an equal number of salient counter-arguments about why the polls are far too favorable for Harris. In the past 4 years or so we've seen polls at variance with results swinging widely either way - either far too or less favorable for either Democrats or Republicans.

My sense is that this race is too close to call either way, although I think that Harris has a slight advantage. I would not however be surprised to see Trump win by a slight margin, or Harris win by a slight or better-than slight margin.

Complicating matters enormously is the Electoral College effect - where the popular vote (which I'm 90% certain will go for Harris) doesn't matter - instead it's the EC that counts - likely down to a very few voters in a very few states. Who-the-fuck knows how that will go?

We'll see.
Click to expand...
The best non biassed pollster/predictor is 338 Canada where highly sophisticated statistical methods are used. It has done some work on US election and currently has Harris 8 points ahead nationally. It is suggesting that EC vote will be close but with a win for Harris 272 votes.
PS EDIT. Incidently, my prediction a long time ago that TEXAS maybe in play is now looking not so outrageous.
VOTE BLUE.
 
Last edited:
KS55 said:
The best non biassed pollster/predictor is 338 Canada where highly sophisticated statistical methods are used. It has done some work on US election and currently has Harris 8 points ahead nationally. It is suggesting that EC vote will be close but with a win for Harris 272 votes.
Click to expand...
Cheers.
338 on my Christmas Card list if Harris wins :-).
 
You must log in or register to reply here.
Unread Posts

Don't have an account? Register now and see fewer ads!

SIGN UP
Back
Top
✖
  AdBlock Detected
Bluemoon relies on advertising to pay our hosting fees. Please support the site by disabling your ad blocking software to help keep the forum sustainable. Thanks.