2024 Pre-Election Political Polls | POLL - Trump would have had 7 point lead over Biden

  • Thread starter Thread starter nycfan
  • Start date Start date
  • Replies: 6K
  • Views: 144K
  • Politics 
Paine said:
Color me shocked, absolutely shocked, that the final NYT poll is tied. Wonder what incentive they have for that.

I’m at the point where I’m just saying fuck it until Election Day. None of these pollsters know what the turnout is going to be. They’re shooting in the dark and hoping they finally got it right this year.
Click to expand...
The NYT was not putting its thumb on the scale for Trump in this poll. They have no incentive to do that. Nate Cohn is well respected. He could get a job anywhere. He's not going to throw his reputation away.

A tied race nationally means almost nothing right now. The forecast models barely even use the national polls at this stage, when there are so many swing state polls.
 
superrific said:
The NYT was not putting its thumb on the scale for Trump in this poll. They have no incentive to do that. Nate Cohn is well respected. He could get a job anywhere. He's not going to throw his reputation away.

A tied race nationally means almost nothing right now. The forecast models barely even use the national polls at this stage, when there are so many swing state polls.
Click to expand...
I’m not saying they’re “putting their thumb on the scale.” It’s well documented that the NYT poll is weighing their poll based on assumptions about a change in dynamics.

That is, they expect Trump to improve on his numbers with low-propensity Black and Latino (and to a lesser extent, White) men. At the same time, they expect Harris to remain strong with college educated whites in the way that Biden did and the party did in 2022.

It remains to be seen whether this will happen or not. Not every poll is doing the same things the NYT poll is doing, and it’s a deliberate choice they’re making.

Even if Cohn thoroughly agrees with the methodology, the fact remains: this drives clicks and attention. Cohn approving the results and methodology is just the icing on the cake for the NYT’s business model.
 
Paine said:
I’m not saying they’re “putting their thumb on the scale.” It’s well documented that the NYT poll is weighing their poll based on assumptions about a change in dynamics.

That is, they expect Trump to improve on his numbers with low-propensity Black and Latino (and to a lesser extent, White) men. At the same time, they expect Harris to remain strong with college educated whites in the way that Biden did and the party did in 2022.

It remains to be seen whether this will happen or not. Not every poll is doing the same things the NYT poll is doing, and it’s a deliberate choice they’re making.

Even if Cohn thoroughly agrees with the methodology, the fact remains: this drives clicks and attention. Cohn approving the results and methodology is just the icing on the cake for the NYT’s business model.
Click to expand...
No, I think you have misunderstood the discussion of the Times' polls. In particular, they aren't weighing their polls on assumptions like that. They are doing the opposite. They are the ones NOT weighing their polls based on assumptions.

Let's review. Every pollster needs a way of creating a representative sample. We talked a bit about Bullfinch and its novel (or perhaps atavistic) approach, which is not to weight at all but call people until they get enough respondents to create the sample they want. Fine, but that's expensive and most pollsters aren't doing it. They are all weighting. They weight by race, gender, age, education, etc.

The point of disagreement is whether to weight on recalled 2020 vote. They are not doing that, unlike many pollsters. I agree with them. Weighting on recalled vote is methodologically suspect and I think it's being used ad hoc by pollsters who are basically looking to add two or three points to Trump in order to be more "accurate." Some pollsters last cycle were just adding 5 points to Trump's numbers, for no reason except they thought Trump was doing better than the polls said. Those pollsters were "accurate" in 2020. The weighting by recalled vote is a more sophisticated version of that.

Anyway, what the recalled-vote weighting does is make an assumption that the electorate is basically the same as in 2020. THAT's THE ASSUMPTION. It's not the Times assuming change; it's the Times being open to the possibility of change.

Keep in mind that the Times polls have given Kamala some great numbers. Great, in the sense of being positive where it matters. And the Times has also been saying, "look, we keep seeing things like Trump +10 in Florida. Other pollsters are massaging that to Trump +5, but we're going to go with Trump +10 because that's what people are telling us." And Trump +10 in Florida is fucking great. If Trump wins Florida by 30, then it will be a Kamala landslide.
 
superrific said:
No, I think you have misunderstood the discussion of the Times' polls. In particular, they aren't weighing their polls on assumptions like that. They are doing the opposite. They are the ones NOT weighing their polls based on assumptions.

Let's review. Every pollster needs a way of creating a representative sample. We talked a bit about Bullfinch and its novel (or perhaps atavistic) approach, which is not to weight at all but call people until they get enough respondents to create the sample they want. Fine, but that's expensive and most pollsters aren't doing it. They are all weighting. They weight by race, gender, age, education, etc.

The point of disagreement is whether to weight on recalled 2020 vote. They are not doing that, unlike many pollsters. I agree with them. Weighting on recalled vote is methodologically suspect and I think it's being used ad hoc by pollsters who are basically looking to add two or three points to Trump in order to be more "accurate." Some pollsters last cycle were just adding 5 points to Trump's numbers, for no reason except they thought Trump was doing better than the polls said. Those pollsters were "accurate" in 2020. The weighting by recalled vote is a more sophisticated version of that.

Anyway, what the recalled-vote weighting does is make an assumption that the electorate is basically the same as in 2020. THAT's THE ASSUMPTION. It's not the Times assuming change; it's the Times being open to the possibility of change.

Keep in mind that the Times polls have given Kamala some great numbers. Great, in the sense of being positive where it matters. And the Times has also been saying, "look, we keep seeing things like Trump +10 in Florida. Other pollsters are massaging that to Trump +5, but we're going to go with Trump +10 because that's what people are telling us." And Trump +10 in Florida is fucking great. If Trump wins Florida by 30, then it will be a Kamala landslide.
Click to expand...
Okay, that makes sense. Thanks for the explanation. Regardless, we’ll have our answer soon.
 
nycfan said:
Alicia Silverstone My Bad GIF
Click to expand...
I really thought you were older than this. And I didn't picture you as a blonde either. I can only imagine what your clients must think when they're picturing some gruff old dude doing their transactional work and this young woman enters the room instead.
 
"... [Harry Enten on CNN]:

"History tells us that it is more likely than not that all of the swing state polling errors would move in one direction," the polling expert said. "So in 2012, 92% of the states moved in Obama‘s direction."

"In 2016, 83% of the swing states move together because the polls underdid Donald Trump, of course we all remember that. And how about in 2020? 100% of the swing state poll averages underestimated Donald Trump, and so he did better than a lot of folks thought," he said.

"So this time around, don‘t be surprised at the swing-state polls when they underestimate one candidate, they underestimate all of them in the states, and that would lead to a relative Electoral College blowout with of one of the candidates winning at least 300 electoral votes," he concluded. ..."

 
Interesting data from NC this week. White people voted in force. On Tuesday, the racial turnout gap was less than 4 points. Now it's more than 7. Not sure if that's meaningful, but in 2020 the gap was 12%. I don't know if black people tend to vote more commonly on weekends, so maybe that's what the data is showing this week.
 
You must log in or register to reply here.
Back
Top