Old McDonald said:

---

it's other factors too but that's the biggest one. PA is the likeliest tipping point state in the model so stands to reason the election odds are sensitive to recent polls tipping from trump to harris.

---

Yet Trump is leading or tied or down by 1-2 points…. So why the shift?!!! PA being the reason for the shift absolutely doesn't make sense. If anything, PA has moved in Trumps favor the last couple of weeks.

The Bullfinch Group put out a poll this week that had Harris within 1 that's sparked a lot of th Harris-Florida talk. I don't necessarily believe that, but I don't believe it's a Trump blowout. Probably like a 2-4 point lead seems more likely given past presidential election results.

https://www.newsweek.com/florida-polling-dead-heat-kamala-harris-donald-trump-1958985

Quote:

---

Kamala Harris and Donald Trump are in a dead heat in Florida, according to a new poll.

The poll, conducted by The Independent Center and The Bullfinch Group between September 20 and 23, found that Trump had a 1-point lead among registered voters in Florida, on 48 percent to Harris' 47 percenta lead well within the poll's 4-point margin of error. The lead was the same when third-party candidates were offered as a choice.

---

Quote:

---

Morning Consult's most recent poll, conducted between September 9 and 18 among 2,948 likely voters, gave Trump a 3-point lead over Harris. Another poll conducted by Morning Consult the week before gave the former president a lead of 2 points. The polls had a margin of error of +/-2 percentage points.

A poll conducted by Florida Atlantic University, Public Opinion Research Lab (PolCom Lab) and Mainstreet Research USA between August 10 and 11 gave Trump a lead of 2 to 3 points among 1,040 likely voters, with a margin of error of +/- 3 percent.

---

Quote:

---

Every poll conducted in the state since Harris entered the race on July 21 has shown Trump leading by 1 to 8 points, according to FiveThirtyEight. Before Harris entered the race, Trump had been leading by between 4 and 13 points.

The most recent Redfield and Wilton Strategies poll, conducted between September 16 and 19 put Trump ahead by 5 points among 1,602 likely votersa lead outside of the margin of error.
...
In 2020, Trump won the state by 3.3 points, up from his 1.2-point victory in 2016.

---

this is probably my single favorite polling factoid I've seen. this tweet is from before the 2020 election but it held true that year, the only swing states where trump or biden polled over 49% were biden in michigan and wisconsin where he ended up winning both.

if the pattern holds true this year, based on current averages harris is beginning to look safe in MI/WI and trump in AZ/GA/NC. PA remains the tipping point.

1/2 Since 2004, a candidate has polled at or above 49% in a battleground state and gone on to lose that state once. One time. #ElectionTwitter

Never in recent history (curious if ever?) - has a candidate polled over 50% and lost.@FiveThirtyEight@RealClearNews pic.twitter.com/bw42MoP3Nr

— Carl Allen (@RealCarlAllen) October 28, 2020

Hillary looked safe in MI/WI and Biden won WI by a printer cartridge.

Trump outperforms the polls until someone proves to me he doesnt.

49% in the final RCP average is the key here. hillary was not above 49% in WI or MI. and biden still won WI.

How is Harris looking good here?

Barnyard96 said:

---

How is Harris looking good here?

---

Because she's winning.

No, that's a statistical tie.

You're also ignoring the fact that HRC lost it with an RCP average at 4% and Biden barely won with a 6% average. Harris is at 1%. If history is any indication about how inaccurate the polls are, she's in a bad spot.

[We've been clear this is a discussion on polls and does not include your personal projection of who's going to win or lose certain states simply based on your opinion. Bring facts, data, or well-informed perspective to the table if you want to add context to the polls. We're done simply removing your derails -- Staff]

[Another ban for derailing from a discussion on polls. We've been clear. More incoming for posters who persist -- Staff]

📊 Swing States Polling: @atlas_intel

PENNSYLVANIA
🟥 Trump: 51% (+2.9)
🟦 Harris: 48.1%
—
WISCONSIN
🟥 Trump: 49.7% (+1.5)
🟦 Harris: 48.2%
—
MICHIGAN
🟥 Trump: 50.6% (+3.4)
🟦 Harris: 47.2%

GEORGIA
🟥 Trump: 49.6% (+0.6)
🟦 Harris: 49%
—
ARIZONA
🟥 Trump: 49.8% (+1.2)… pic.twitter.com/eYNGsvqlYE

— InteractivePolls (@IAPolls2022) September 29, 2024

Good numbers for Trump. North Carolina outlier

Quo Vadis? said:

---

📊 Swing States Polling: @atlas_intel

PENNSYLVANIA
🟥 Trump: 51% (+2.9)
🟦 Harris: 48.1%
—
WISCONSIN
🟥 Trump: 49.7% (+1.5)
🟦 Harris: 48.2%
—
MICHIGAN
🟥 Trump: 50.6% (+3.4)
🟦 Harris: 47.2%

GEORGIA
🟥 Trump: 49.6% (+0.6)
🟦 Harris: 49%
—
ARIZONA
🟥 Trump: 49.8% (+1.2)… pic.twitter.com/eYNGsvqlYE

— InteractivePolls (@IAPolls2022) September 29, 2024

Good numbers for Trump. North Carolina outlier

---

I've been saying it for a while, the election hinges on PA.

Silvers numbers for kamala mirror her PA chances. If Trump takes PA it's over

Atlas was the best pollster in 2020. Interesting he is doing so well in the Rust Belt but trialing in NC and razor thin in Georgia.

Their senate polls

📊 Senate polling by @atlas_intel

MICHIGAN
🟥 Rogers: 49.3% (+5.7)
🟦 Slotkin: 43.6%
—
ARIZONA
🟦 Gallego: 49.8% (+4.1)
🟥 Lake: 45.7%
—
NEVADA
🟦 Rosen (inc): 47.5% (+1.5)
🟥 Brown: 46%
—
WISCONSIN
🟦 Baldwin (inc): 48.2% (+1.5)
🟥 Hovde: 46.7%
—
PENNSYLVANIA
🟦 Casey…

— InteractivePolls (@IAPolls2022) September 29, 2024

Captn_Ag05 said:

---

Atlas was the best pollster in 2020. Interesting he is doing so well in the Rust Belt but trialing in NC and razor thin in Georgia.

---

It would be interesting if Silver would adjust his model ... weighting polls based on previous years accuracy.

Someone earlier mentioned that Silver double counts Morning Consult. Is that due to him counting a Morning Consult poll as well as another labeled poll that's actually performed by Morning Consult?

Barnyard96 said:

---

How is Harris looking good here?

---

because she's at 49%, which if you refer back to my original post is the singular basis of the predictive theory

If you start ignoring these ridiculous Bloomberg polls, that number drops below 49. It's almost like they are doing this to keep the aggregates close

Captn_Ag05 said:

---

Atlas was the best pollster in 2020. Interesting he is doing so well in the Rust Belt but trialing in NC and razor thin in Georgia.

---

Black support breaking harder for Harris? Idk.

the theory is just 49% or higher in the RCP average, no caveats for "49% but exclude polls you dont like". it's just a theory, always possible it won't hold up! but the idea is once you cross 49% support, it's just mathematically hard for your opponent to pick up more than that from what's left.

I actually don't disagree with that in theory and I would feel better if she was at 47 and no higher. Still, the only thing keeping Trump from having 270+ on RCP was Bloomberg. It's justified to ignore clearly biased polls just like I take Rasmussen with a grain of salt.

a lot of people forget that Hillary might have had leads in the late MI/WI/PA polls, but her support was capped around like 46%. lots of "undecideds" left over for trump to pick up, and he did. that's not so much the case with the polling this election.

Well Trump is at 49% in some of those WI polls too.

dreyOO said:

---

Captn_Ag05 said:

---

Atlas was the best pollster in 2020. Interesting he is doing so well in the Rust Belt but trialing in NC and razor thin in Georgia.

---

It would be interesting if Silver would adjust his model ... weighting polls based on previous years accuracy.

---

He does.

2004FIGHTINTXAG said:

---

Not buying a 16 point swing in Nate Silver's model based on a less than 1% difference in PA.

---

It isn't just due to that. Silver's code is not public, so can't be precise with exactly where shifts come from, but probably the largest contributor was the expiration of a "convention bounce adjustment" for Harris.

The model "expects" certain things to happen based on historical precedent, for example, presidential candidates have typically seen about a +4 increase in polls in the week or two after the candidate's party convention. Since polls for Harris only went up by 1 or 2 points, the model sees that as an underperformance and adjusts her odds accordingly.

You could argue (as a lot of people on Team D did on Twitter and Silver's website) that the convention bounce probably doesn't exist any more to any meaningful degree. In any event, unless you were wagering on the election during the weeks in question it doesn't matter, as the "bounce" is considered a temporary event in any case.

A "polls only" version of the model would have had Harris as a very narrow favorite over the past month, but clearly in toss-up range the whole time.

Nice, more reinforcement about this shift...

🚨Holy sht!

Trump has gained 19 points since 2020 among Latino voters…!

This is how you win elections! pic.twitter.com/4GbdrIN9ew

— Gunther Eagleman™ (@GuntherEagleman) September 29, 2024

Fox News showing this much of a lead for Trump vs. the Border Tsar on the border is pretty funny;

NEW POLL: President Trump holds a 15-point lead among Arizona voters on trust to handle the border pic.twitter.com/hwDrLBw6GI

— Trump War Room (@TrumpWarRoom) September 29, 2024

Back to the Silver discussion, if this is true it really is stupefying why he is doing this; Big Village and Quinnipiac in the top 3??? LOL, no wonder he's confident in Kackling Kamala right now.

Yo @Peoples_Pundit check this out.

Nate puts 37% more "influence" on Quinnipiac than us. pic.twitter.com/Itmu2r0Gkr

— Mark Mitchell, Rasmussen Reports (@Mark_R_Mitchell) September 27, 2024

Any idea how Nate is funded?

Barnyard96 said:

---

Well Trump is at 49% in some of those WI polls too.

---

again, the theory's track record is only valid using the final RCP average. trump is not at 49% in the current RCP average for WI and harris is, but that can always change.

So I need some help understanding this data:

Gallup shows voters identifying themselves GOP +4 over Dems
Trump getting ~ 40% of Hispanic votes
Trump getting ~ 20% of the black votes
Trump getting ~54% of the teamsters vote

If those polls that indicate this are right, how on earth is it even close?

nortex97 said:

---

Back to the Silver discussion, if this is true it really is stupefying why he is doing this; Big Village and Quinnipiac in the top 3??? LOL, no wonder he's confident in Kackling Kamala right now.

---

here's a link to useful reading about how silver's model assigns "influence" and adjusts for statistical partisan bias across polls. the short answer is "influence" is tied to factors related to statistical quality (i.e. large sample size) and they do apply house effect adjustments (if a poll statistically tends to favor Rs or Ds the model will apply a corrective adjustment to it).

good to read if you've ever found yourself upset by silver's inclusion of certain polls.

It won't make much difference given that Rasmussen already has a strong R-leaning house effect (see below) but still this is the right precedent to set per our standards.https://t.co/xf1EIoWTcl

— Nate Silver (@NateSilver538) September 28, 2024

If coordination with a campaign results in being labeled as an R, then NYTimes/NBC/ABC/CNN polls all need a big fat D next to them

[You're re-plowing ground that's already been explained. Stop derailing the thread with it -- Staff]

I appreciate it but that all doesn't make much sense at all to me. Rasmussen maybe they are using an algorithm for R Bias but they were most accurate in 2016-2020. Having an algorithm doesn't mean the algorithm is right.

And the new republic article was frankly absurd. Rasmussen reached out to a 501c3 for funding/sponsorship? Ok. All the pollsters try to get paid.

More good news (from elsewhere but so it is here as well):

NBC News/ Telemundo poll of Latino voters

🟦 Harris 54% (+14)
🟥 Trump 40%

1,000 RV | 9/16-9/23

2020
🟦 Biden 63% (+36)
🟥 Trump 27%

2016
🟦 Clinton 69% (+50)
🟥 Trump 19%

— David D. Chapman (@davidchapman141) September 29, 2024

Driven by men/younger Latino's:

What's behind this Latino-voter erosion for Democrats? Men and younger Latinos pic.twitter.com/AhsLLHU6wD

— Mark Murray (@mmurraypolitics) September 29, 2024

i trust AtlasIntel more than his model now

— lxx (📜, ⚡️) (@luoxiaxia) September 29, 2024

If its all down to PA, and the polls say its a dead heat, what you guys thinking Trump's odd are here?

Does he over-perform like the last 2 elections?