FL-NBC/Marist: TIE

[Pollster]

Note that this poll is weighted by income rather than education, which is correlated in many ways but distinct in many others.

[wbrocks67]

The poll has a 47-43 approval rating amongst Latino voters. CIVIQS have it 34-64. It could just be a Trump heavy sample that influences the poll.

Yeah, Civiqs is showing a way better showing for Dems with Latinos too. The GCB is like 71-27 for Dems. Makes me wonder if like we've talked about, that polling is just really terrible with Hispanics. Just as it's been for years now.

[Sprouts Farmers Market ✘]

Demographics are destiny 😍

[wbrocks67]

Note that this poll is weighted by income rather than education, which is correlated in many ways but distinct in many others.

They're still not weighting by education??

[Fmr. Gov. NickG]

35% GOP
32% Dem
31% Indie

Um...did they oversample Republicans?

It's hard to know what the party ID of the population actually should be.

But the 38/36/26 party registration split of the RV sample seems right on the money.

[Alben Barkley]

It's usually a bad idea to question shonky margins in crosstabs, although the lack of weighting by education is a more serious concern.

Lean D -> Tilt D, but deep down we all know this is an immutably tilt R state.

Changing rating because of one poll is ridiculous. As GeorgiaModerate rightly pointed out, if Biden actually has a modest lead here, a tie and polls with larger leads are both to be expected. But if this is the low end of his range and the best Trump can get, Biden is still clearly favored.

[afleitch]

I'm not really a huge burrower into crosstabs, but pretty much everything in this poll is so counter-intuitive. Even at that it still gives a tie.

[wbrocks67]

It's usually a bad idea to question shonky margins in crosstabs, although the lack of weighting by education is a more serious concern.

Lean D -> Tilt D, but deep down we all know this is an immutably tilt R state.

Changing rating because of one poll is ridiculous. As GeorgiaModerate rightly pointed out, if Biden actually has a modest lead here, a tie and polls with larger leads are both to be expected. But if this is the low end of his range and the best Trump can get, Biden is still clearly favored.

Not to mention, Biden being *tied* among <45 year olds and LOSING Hispanics just doesn't jive with reality.

[TiltsAreUnderrated]

It's usually a bad idea to question shonky margins in crosstabs, although the lack of weighting by education is a more serious concern.

Lean D -> Tilt D, but deep down we all know this is an immutably tilt R state.

Changing rating because of one poll is ridiculous. As GeorgiaModerate rightly pointed out, if Biden actually has a modest lead here, a tie and polls with larger leads are both to be expected. But if this is the low end of his range and the best Trump can get, Biden is still clearly favored.

It's not ridiculous with a poll of this quality and this close to election day (negative marks for not weighting by education, but otherwise fine). FL was a weak lean D in my book and now it is a strong tilt D.

[GeorgiaModerate]

FL just flipped red on PredictIt.

I'll take Overreactions to a Single Poll for $1000, Alex.

[Arizona Iced Tea]

Look at this beautiful tightening! Biden's once 8 point lead now just 1.4

[BigSerg]

Trump +1

https://twitter.com/Politics_Polls/status/1303380627434151937?s=19

[Alben Barkley]

It's usually a bad idea to question shonky margins in crosstabs, although the lack of weighting by education is a more serious concern.

Lean D -> Tilt D, but deep down we all know this is an immutably tilt R state.

Changing rating because of one poll is ridiculous. As GeorgiaModerate rightly pointed out, if Biden actually has a modest lead here, a tie and polls with larger leads are both to be expected. But if this is the low end of his range and the best Trump can get, Biden is still clearly favored.

It's not ridiculous with a poll of this quality and this close to election day (negative marks for not weighting by education, but otherwise fine). FL was a weak lean D in my book and now it is a strong tilt D.

1. The crosstabs and weighting make the poll’s quality suspect. Even quality pollsters sometimes have outliers and just bad polls.

2. Because of the fact that even quality pollsters are not infallible, and also for the reasons I stated which you completely failed to address (expected range, margin of error, etc.), it is ALWAYS wrong to overreact to ANY one poll from ANY pollster. Period. Unless and until more polls come out that in aggregate suggest Biden’s overall lead has declined here — hell, when was the last time Trump even led here, which he definitely would in some if this race was truly this close? — changing your overall view of the state of the race is a foolish, ridiculous, irrational kneejerk reaction. End of story.

[SenatorCouzens]

Doesn't not weighting polls by education usually mean the poll is bias in favor of Democrats? Because whites with college degree are much more likely to respond to polls and much more likely to vote for Democrats, compared to whites without a college degree.

Or am I wrong about that generally? Or is it correct but Florida is different?

[Dr Oz Lost Party!]

FL just flipped red on PredictIt.

I'll take Overreactions to a Single Poll for $1000, Alex.

It's hilarious how multiple good polls for Biden in a state like North Carolina won't even move the needle on PredictIt, but one bad poll will knock Biden down 7-10 points. Texas is still at 73 for Trump despite Biden leading in several polls there. That just shows how Trump-friendly the betting markets are.

[Roronoa D. Law]

Forget Puerto Rico. Can we give FL back to Spain free of charge?

[gf20202]

Doesn't not weighting polls by education usually mean the poll is bias in favor of Democrats? Because whites with college degree are much more likely to respond to polls and much more likely to vote for Democrats, compared to whites without a college degree.

Or am I wrong about that generally? Or is it correct but Florida is different?

That's not what weighting by education means. It doesn't mean having more people in the poll who have gone to college

[TiltsAreUnderrated]

It's usually a bad idea to question shonky margins in crosstabs, although the lack of weighting by education is a more serious concern.

Lean D -> Tilt D, but deep down we all know this is an immutably tilt R state.

Changing rating because of one poll is ridiculous. As GeorgiaModerate rightly pointed out, if Biden actually has a modest lead here, a tie and polls with larger leads are both to be expected. But if this is the low end of his range and the best Trump can get, Biden is still clearly favored.

It's not ridiculous with a poll of this quality and this close to election day (negative marks for not weighting by education, but otherwise fine). FL was a weak lean D in my book and now it is a strong tilt D.

1. The crosstabs and weighting make the poll’s quality suspect. Even quality pollsters sometimes have outliers and just bad polls.

Agreed on the lack of education weighting, but the crosstabs do not make the poll's quality particularly suspect. Reading too far into them - especially with a full sample size that isn't in the thousands - is not far from auguring from chicken guts, beyond the most absolutely ridiculous results (e.g. Trump winning black voters while losing the topline).

As we move closer to e-day, I am going to rely more on polls than fundamentals. FL hasn't had results from another A-rated pollster since June, so this is going to be more important than most of the recent ones for me.

2. Because of the fact that even quality pollsters are not infallible, and also for the reasons I stated which you completely failed to address (expected range, margin of error, etc.), it is ALWAYS wrong to overreact to ANY one poll from ANY pollster. Period. Unless and until more polls come out that in aggregate suggest Biden’s overall lead has declined here — hell, when was the last time Trump even led here, which he definitely would in some if this race was truly this close? — changing your overall view of the state of the race is a foolish, ridiculous, irrational kneejerk reaction. End of story.

It is a small percentage change in my view of the race. My ratings are probablistic and I now think Biden's chance of winning FL is slightly below 59% as opposed to slightly over 59%. That's all it takes to move a race from lean to tilt.

Trump last led here from September 1-3 according to Trafalgar; although that is an extremely trashy poll and I treat it as a GOP internal, the moronic social desirability weighting isn't new (so they aren't guaranteed to be fraudulent). Prior to that, he got ties in several polls in June and July; these were pretty trashy, but the better ones showing Biden ahead weren't of Marist's standard and most showed Trump within the margin of error.

[American2020]

https://twitter.com/LarrySabato/status/1303388514453721092

[TiltsAreUnderrated]

https://twitter.com/LarrySabato/status/1303388514453721092

"Please let the horse race narrative be true."

You can almost hear Cook getting ready to paint everything as a tossup.

[Crumpets]

Florida gonna Florida.

[SenatorCouzens]

Doesn't not weighting polls by education usually mean the poll is bias in favor of Democrats? Because whites with college degree are much more likely to respond to polls and much more likely to vote for Democrats, compared to whites without a college degree.

Or am I wrong about that generally? Or is it correct but Florida is different?

That's not what weighting by education means. It doesn't mean having more people in the poll who have gone to college

Right... but because whites with college are invariably more likely to respond to polls than whites without college, not accounting for education in polls is almost always going to oversample those with college degrees?

[Devils30]

We need Nate Cohn to do a poll here.

[Buzz]

Marist thought Gillum was going to win the state by 4, and don’t get me started on Senator Bill Nelson LOL

[Woody]

Enjoy the retirement home Joe!