The Upshot on Today’s Polls
Analysis of the day in polls, from now until Election Day.
| Day | Place | Result |
|---|---|---|
| Tuesday | Nevada | Biden +6 |
| Wednesday | Michigan | Biden +8 |
| Thursday | N.C. | Biden +3 |
| Sunday | Arizona | Biden +6 |
| Sunday | Florida | Biden +3 |
| Sunday | Pennsylvania | Biden +6 |
| Sunday | Wisconsin | Biden +11 |
| Polling leader | If polls are as wrong as they were in… | ||||
|---|---|---|---|---|---|
| 2016 | 2012 | ||||
| U.S. | +8 Biden | +6 | +12 | ||
| N.H. | +11 Biden | +7 | +14 | ||
| Wis. | +10 Biden | +4 | +14 | ||
| Minn. | +10 Biden | +4 | +12 | ||
| Mich. | +8 Biden | +4 | +13 | ||
| Pa. | +6 Biden | +1 | +7 | ||
| Nev. | +6 Biden | +8 | +9 | ||
| Neb. 2* | +5 Biden | +9 | <1 | ||
| Maine 2* | +3 Biden | +9 | +8 | ||
| Ariz. | +3 Biden | +1 | +1 | ||
| Fla. | +2 Biden | <1 | +4 | ||
| N.C. | +2 Biden | +3 | +3 | ||
| Ga. | +2 Biden | <1 | +2 | ||
| Ohio | <1 Trump | +6 | <1 | ||
| Iowa | +1 Trump | +6 | +3 | ||
| Texas | +2 Trump | +4 | +1 | ||
A polling error much bigger than 2016’s.
President Trump’s surprising victory in 2016 gave him an aura of political invincibility. But he’s in a far bigger predicament now than the one he faced heading into Election Day in 2016. The polls show Joe Biden with a far more significant lead than the one held by Hillary Clinton, and many of the likeliest explanations for the polling misfire do not appear to be in play today.
Over at The Upshot, I go into the ways the polls are different today than they were in 2016, and what sort of polling error Mr. Trump would need to win at this point. (A big one.)
What Trump Needs to Win: A Polling Error Much Bigger Than 2016’s »
November 2
A running table of all the presidential polls released today, with newest entries first:
Advertisement
Online polls or automated phone polls, the lead stays steady.
Another 20 polls today, another 20 polls without signs of gains for the president.
State polls
National polls
| U.S. |
Léger
Oct. 29-Nov. 1, 827 L.V.
|
Biden +8
50-42
|
+6D |
| U.S. |
IBD/TIPP
Oct. 27-31, 1,072 L.V.
|
Biden +5
49-44
|
+3D |
| U.S. |
Data for Progress (Dem. pollster)
Oct. 28-29, 1,403 L.V.
|
Biden +10
54-44
|
+8D |
It’s not just the live phone polls. This morning, a big wave of polls showed Joe Biden with a clear lead. Nearly all of these were sponsored by major media organizations and conducted via live phone interviews. If you want to check out what we had to say about them, you can scroll down a bit farther on the page. TLDR: Biden’s still ahead by a lot.
This afternoon, we got a big wave of polls showing a very similar result, but by very different means. Nearly all were conducted online or by automated phone calls. Emerson, Ipsos, YouGov (here as part of the Cooperative Election Study) and SurveyMonkey are all fairly familiar pollsters with this approach.
These polls differ from the live phone polls in several respects. Most obviously, they don’t require respondents to divulge their attitudes to a human, which ought to make the respondents likelier to voice an unpopular view. If there really are shy Trump supporters, this is where you would think they would finally feel free to speak out.
Less obviously: The online and automated firms usually take different strategies for ironing out bias, like weighting on party identification or on whether respondents said they voted for President Trump or Hillary Clinton in 2016.
I don’t want to get into the ins and outs of which methods might be best. What’s important: Mr. Biden seems to hold a significant lead, no matter the methodology. It suggests that his lead doesn’t hinge on, say, a certain set of choices common to the major pollsters. It also suggests that his lead doesn’t instantly vanish once you stop using live phone calls and the shy Trump voters start getting real.
The Trump pollsters are all alone. There are some polls showing Mr. Trump ahead today, and they’re mainly from a familiar quartet: Trafalgar, Susquehanna Research, Insider Advantage and Rasmussen.
| Pollster | Margin | Diff. from ’16 result | |
|---|---|---|---|
| Fla. |
Susquehanna Polling & Research Inc.
Oct. 29-Nov. 1, 400 L.V.
|
Trump +1
46-47
|
<1D |
| Fla. |
Rasmussen Reports/
Pulse Opinion Research Oct. 29-31, 800 L.V.
|
Biden +1
48-47
|
+2D |
| Ga. |
Opinion Savvy/
InsiderAdvantage Nov. 1-1, 500 L.V.
|
Trump +2
46-48
|
+3D |
| Iowa |
Opinion Savvy/
InsiderAdvantage Oct. 30-30, 400 L.V.
|
Trump +2
46-48
|
+7D |
| Mich. |
Opinion Savvy/
InsiderAdvantage Oct. 30-31, 500 L.V.
|
Biden +2
49-47
|
+2D |
| N.C. |
Opinion Savvy/
InsiderAdvantage Oct. 30-31, 450 L.V.
|
Trump +4
44-48
|
<1R |
| N.C. |
Trafalgar Group
Oct. 27-29, 1,082 L.V.
|
Trump +2
47-49
|
+2D |
| Pa. |
Opinion Savvy/
InsiderAdvantage Oct. 30-31, 500 L.V.
|
Trump +2
47-49
|
+1R |
These are also automated or partly online surveys. But they differ from the other online and automated surveys in an important respect: They’re tough to distinguish from partisan pollsters. They’re generally not very transparent, they’re vocal about their support for the president, and they often have partisan sponsors. The Susquehanna and Insider Advantage polls are sponsored by the Center for American Greatness, which has a website that leads with an article comparing this election and 1776. Rasmussen and Trafalgar will undoubtedly be familiar to most anyone reading this page. For the uninitiated: These results are no surprise.
It is still possible that Mr. Trump will ultimately win this election. If he does, some of these pollsters will probably say it vindicates their methodology. I wouldn’t say so. There’s nothing about their methodology, as described, that explains how they arrive at results so different from other pollsters using seemingly similar means — including those described earlier in this post. The easier explanation is that these pro-Trump pollsters get pro-Trump results because they’re pro-Trump, not because they and they alone have unlocked the secrets to public polling.
But take note: Even these polls show Mr. Biden and Mr. Trump essentially tied in Florida, with Mr. Biden leading in Michigan, and with the president underperforming his standing from four years ago. If the tie went Mr. Biden’s way in Florida, that would all but certainly be enough to make him president.
The president is just about out of time.
The final polls are beginning to arrive. They don’t bring good news for the president.
State polls
| Pollster | Margin | Diff. from ’16 result | |
|---|---|---|---|
| Ariz. |
New York Times/
Siena College Oct. 26-30, 1,252 L.V.
|
Biden +6
49-43
|
+10D |
| Fla. |
New York Times/
Siena College Oct. 27-31, 1,451 L.V.
|
Biden +3
47-44
|
+4D |
| Fla. |
St. Pete Polls
Oct. 29-30, 2,758 L.V.
|
Biden +1
49-48
|
+2D |
| Fla. |
RMG Research
Oct. 28-30, 1,200 L.V.
|
Biden +4
51-47
|
+5D |
| Fla. |
ABC News/
The Washington Post Oct. 24-29, 824 L.V.
|
Trump +2
48-50
|
<1R |
| Iowa |
Emerson College
Oct. 29-31, 604 L.V.
|
Trump +2
47-49
|
+7D |
| Mich. |
Emerson College
Oct. 29-31, 700 L.V.
|
Biden +6
52-46
|
+6D |
| Mich. |
EPIC-MRA
Oct. 25-28, 600 L.V.
|
Biden +7
48-41
|
+7D |
| N.M. |
Research & Polling Inc.
Oct. 23-29, 1,180 L.V.
|
Biden +12
54-42
|
+4D |
| Ohio |
Emerson College
Oct. 29-31, 656 L.V.
|
Biden +1
50-49
|
+9D |
| Pa. |
New York Times/
Siena College Oct. 26-31, 1,862 L.V.
|
Biden +6
49-43
|
+7D |
| Pa. |
ABC News/
The Washington Post Oct. 24-29, 810 L.V.
|
Biden +7
51-44
|
+8D |
| Wis. |
New York Times/
Siena College Oct. 26-30, 1,253 L.V.
|
Biden +11
52-41
|
+12D |
National polls
| U.S. |
U.S.C. Dornsife
Oct. 18-31, 5,364 L.V.
|
Biden +12
54-42
|
+10D |
| U.S. |
NBC News/
The Wall Street Journal Oct. 29-31, 1,000 R.V.
|
Biden +10
52-42
|
+8D |
No sign of a shift. Sunday morning’s polls look about the same as everything else we’ve seen over the last two weeks: Joe Biden enjoys a commanding national lead, with a significant if somewhat smaller lead in the battleground states likeliest to decide the election.
If you squint, you could argue Mr. Biden’s lead is down a point compared with the prior iteration of these polls. The NBC/WSJ poll shows him up 10 points, compared with 11 points two weeks ago, for instance. The NYT/Siena polls show Mr. Biden’s lead falling by an average of one point across the four surveys.
But most of these prior surveys were conducted a while ago, and often in the wake of the first presidential debate and President Trump’s coronavirus hospitalization. We already knew that the polls were somewhat tighter this past week than they were in early October, when the president was at his nadir. You can see that for yourself by looking at our poll average charts.
What we don’t see here — or really anywhere — is a sign that the race has tightened over the final stretch compared with, say, a week or so ago.
Biden ahead in Pennsylvania. We’ve been waiting and waiting for the final polls of Pennsylvania, and now they’re finally starting to arrive. So far, they tell a clear story: Mr. Biden still holding a meaningful lead, though it’s a little tighter.
The final Times/Siena poll shows Mr. Biden up by six points in the state, right between his seven-point lead in the ABC/Post poll and his five-point lead in the most recent Muhlenberg College poll. Our average puts Mr. Biden up by 5.7 points in the state.
It’s a significant lead, but there’s still quite a bit of daylight between the polling there and either the national polls or the polling in Michigan and Wisconsin, where today every survey again showed Mr. Biden up by at least seven points — including an 11-point lead in Wisconsin in the final Times/Siena survey.
The Pennsylvania polls are also just a little bit tighter than earlier in the month, when ABC/Post and Times/Siena put Mr. Biden up by wider margins. We’ve seen a similar trend in most of the other telephone polls of Pennsylvania, like Fox and Quinnipiac, which also showed Mr. Biden up by at least eight points in September or October. We’re still waiting for the final Monmouth poll of the state, but I’d guess it’ll show a closer race than its last poll, which had Biden up by 11 points.
As we’ve said over and over, Pennsylvania is the key to whether Mr. Biden is an overwhelming favorite or merely a clear favorite. You might be bored of this formulation by now, but it’s still true: Mr. Biden is clearly ahead in the Clinton states, Nebraska’s Second District and Michigan and Wisconsin. That puts him at 259 electoral votes, or one Arizona-or-larger-sized battleground away from the presidency. Pennsylvania is the best option for Mr. Biden. So if he has a commanding lead in Pennsylvania — comparable to his leads in Wisconsin and Michigan today, or comparable to his leads in Pennsylvania earlier in October — he has a commanding lead in enough states to win.
Mr. Biden’s six-point lead looks solid, but it is short of “commanding.” It’s at the edge of the range where it’s reasonable to imagine an upset. It doesn’t happen often, and it would be a surprise. But it does happen. It happened in Wisconsin in 2016, for instance. If you look at our “if the polls are as wrong as they were four years ago” chart, you can see that Mr. Biden would win Pennsylvania by a point if the polls were off by as much as they were back then. Put differently: The polls wouldn’t need to be that much more off in Pennsylvania for Mr. Trump to carry the state.
Mr. Biden is spending all day in Pennsylvania on Monday, and it’s not hard to see why.
Arizona: Plan B? If Mr. Biden falls short in Pennsylvania, his best backup plan may be Arizona. It’s not a state that has gotten a lot of polling attention this year from the big hitters, but the final Times/Siena poll of the state shows Mr. Biden up by six points, 49 percent to 43 percent. It’s the same as his lead in our poll of Pennsylvania.
Over all, Mr. Biden leads by four points in our Arizona average, so I’m inclined to say that Pennsylvania still represents Plan A for Mr. Biden. But as far as Plan B goes, Arizona’s a decent enough option: It’s worth 11 electoral votes, so it would get Mr. Biden to exactly the 270 he needs to win if he carries those 259 electoral votes that we think sit pretty clearly in his column.
But there’s another reason Arizona is a tantalizing backup plan for the Democrats: It’s not a state with much of a recent history of polling error. The polls missed by only two points or so in 2016, about the same as the national miss. Democrats in the state were actually underestimated by the final polls in the 2018 midterm election. And the state doesn’t have big groups of ancestrally Democratic white working-class and rural voters whom the president might hope to win back over the final stretch.
Florida is Florida. If Mr. Biden’s up eight or nine nationally, you’d sure think he’d figure out how to scratch out a narrow win in Florida … right? Maybe not. The final polls still show a tight race in the perpetually competitive state, and we have a bit of a split decision in the final polling today: NYT/Siena shows Mr. Biden up by three, ABC/Post shows Mr. Trump up by two, and St. Pete Polls (the rare robopoll that really tries to do a good job) shows Mr. Biden up by a point.
Put it all together, and our polling average shows Mr. Biden up by just under two points in the state. Remarkably, that’s closer than our averages in North Carolina, Georgia and Arizona, where Mr. Trump won by a more comfortable margin in each case four years ago.
I just don’t know what to expect on Election Day. The polls were obviously off here in 2016 and 2018, and it’s a legitimately difficult state to poll. It’s so diverse, with so many tiny subgroups — like the state’s mostly white rural northern vote, or its heavily Republican Cuban-American vote in Miami-Dade County — that are easy to miss. And weighting might not make up for it.
Up next. We get Monmouth in Pennsylvania tomorrow. Maybe we’ll get some final state polling out of Fox, CNN/SSRS or NBC/Marist. Maybe we’ll get a final run of Pew data? Or maybe the Cook/KFF/SSRS battleground polling? Maybe a final national poll from ABC/Washington Post? There will probably be some final online surveys from the likes of YouGov, Ipsos and Morning Consult. Regardless, we’re near the end. I don’t expect our averages to move much from here.
Nov. 1
A running table of all the presidential polls released today, with newest entries first:
Advertisement
Biden leads by six points in Pennsylvania and in Arizona; by 11 in Wisconsin; and by three in Florida.
My colleagues Alexander Burns and Jonathan Martin have the story.
New York Times/Siena College poll of likely voters in Pennsylvania
Take this Iowa result seriously, but remember it’s only one poll.
State polls
| Pollster | Margin | Diff. from ’16 result | |
|---|---|---|---|
| Ariz. |
CNN/SSRS
Oct. 23-30, 865 L.V.
|
Biden +4
50-46
|
+8D |
| Ariz. |
Y2 Analytics
Oct. 15-24, 700 L.V.
|
Biden +3
50-47
|
+7D |
| Fla. |
AtlasIntel
Oct. 28-29, 786 L.V.
|
Even
49-49
|
+1D |
| Iowa |
Selzer & Co.
Oct. 26-29, 814 L.V.
|
Trump +7
41-48
|
+2D |
| Mich. |
CNN/SSRS
Oct. 23-30, 907 L.V.
|
Biden +12
53-41
|
+12D |
| Minn. |
P.P.P. (Dem. pollster)
Oct. 29-30, 770 voters
|
Biden +11
54-43
|
+9D |
| N.C. |
CNN/SSRS
Oct. 23-30, 901 L.V.
|
Biden +6
51-45
|
+10D |
| Neb. 2* |
Emerson College
Oct. 29-30, 806 L.V.
|
Biden +2
50-48
|
+4D |
| Pa. |
Muhlenberg College
Oct. 23-28, 419 L.V.
|
Biden +5
49-44
|
+6D |
| Texas |
P.P.P. (Dem. pollster)
Oct. 28-29, 775 voters
|
Biden +2
50-48
|
+11D |
| Wis. |
Susquehanna Polling & Research Inc.
Oct. 29-31, 450 L.V.
|
Biden +3
49-46
|
+4D |
| Wis. |
Emerson College
Oct. 29-30, 751 L.V.
|
Biden +8
53-45
|
+9D |
| Wis. |
CNN/SSRS
Oct. 23-30, 873 L.V.
|
Biden +8
52-44
|
+9D |
National polls
| U.S. |
IBD/TIPP
Oct. 26-30, 1,047 L.V.
|
Biden +6
50-44
|
+4D |
| U.S. |
U.S.C. Dornsife
Oct. 17-30, 5,343 L.V.
|
Biden +12
54-42
|
+10D |
| U.S. |
Gravis Marketing
Oct. 27-29, 1,281 L.V.
|
Biden +6
50-44
|
+4D |
| U.S. |
Angus Reid Global
Oct. 23-28, 2,231 L.V.
|
Biden +8
53-45
|
+6D |
An eye-opening result in Iowa. Let’s start with the poll that has every poll junkie buzzing tonight: the vaunted Des Moines Register/Ann Selzer poll of Iowa. If it’s not President Trump’s best poll of the cycle, I’m not sure what is: Mr. Trump led by seven points, 48 percent to 41 percent, in a state thought to be close enough that Joe Biden visited there yesterday.
Four years ago, the final Selzer poll also showed Mr. Trump leading by seven points (he won by nine). In retrospect, it clearly foreshadowed the president’s late surge among white voters without a degree across the Midwest. Last night, I wrote that “many Biden supporters will be quite unnerved if the Selzer poll again shows the president with surprising strength at the finish.” Based on the text messages I’ve received from Biden supporters tonight, I can report that my prediction was accurate.
The poll is legitimately good news for the president, and it should be taken seriously. But taking a poll seriously does not entail taking it to the bank — not even if it’s the well-regarded Selzer poll. We just had a similar case in the opposite direction: an ABC News/Washington Post poll earlier this week that showed Mr. Biden ahead by 17 points in Wisconsin. When that poll came out, I noted that it should move your view of the race toward Mr. Biden, even though it was clearly an outlier. The same is true of this survey: It should move your view of the race in Iowa toward Mr. Trump. Indeed, he now leads in our Iowa average by two points.
Usually, that would be all that was worth saying. But the Selzer poll is different. It is put on a pedestal. It is the gold standard of the gold standard. So this isn’t a case where I have to remind you to still take note of an outlying result. This is probably a case where I have to go a little further to remind you that no pollster is perfect, and that no pollster can defy the inherent limits of a random sample of 800 voters.
The Selzer poll has an exceptional track record, but any pollster with as many polls under her belt as Ms. Selzer will inevitably wind up with misses — even bad misses. As I put it ahead of the Iowa caucus in 2016: “Of course, even if Ms. Selzer’s methodology was perfect, there’s no reason to expect her results will be. It’s the nature of the game; even a perfectly conducted poll will occasionally yield imperfect results, and sometimes terrible ones.” Her final poll, released that day, showed Mr. Trump leading by five points; Ted Cruz would win by three.
The Selzer poll has quite a bit in common with the ABC/Post poll. Both use random digit dialing (R.D.D.), a technique that allows the pollster to reach a random sample of all adults. In contrast, many state polls, including the Times/Siena poll, call voters from a list of registered voters. Both techniques have their advantages and disadvantages, but the big drawback of R.D.D. polling is that it can be noisy and prone to outlying results.
Why? An R.D.D. pollster can weight only on the demographic characteristics of the full adult population, like age or race. As a result, it has relatively few means to control the partisan balance of its sample. That’s especially true in Iowa, where voters aren’t particularly divided along demographic lines; therefore, demographic weighting doesn’t do much to ensure a politically balanced sample.
In contrast, a pollster using the voter file, like Monmouth or Times/Siena, can ensure a proper balance between registered Democrats and Republicans. That’s not something the R.D.D. pollsters can do, and helps explain why ABC/Post and Selzer surveys were somewhat likelier than average to have the week’s outlying results.
In fact, Ms. Selzer doesn’t weight her polls nearly as much as other R.D.D. pollsters, including the ABC/Post poll. Rather than weight on a bunch of different variables, like population density, education, race and so on, she uses a technique called post-stratification weighting. That’s a fancy way of saying she splits her sample into tiny groups of voters — by age, congressional district and gender (say young men in Iowa’s First district) — and makes sure that each group represents the right share of the sample.
And this year’s Selzer poll looks a lot more like an outlier than the Selzer poll did four years ago. Back then, the Monmouth poll in September showed Mr. Trump up by eight points; this cycle, Monmouth’s poll in October showed Mr. Biden ahead. More broadly, national polling four years ago showed Mr. Trump running very well among white voters without a degree, who are overrepresented in Iowa. This year, polls show Mr. Biden doing far better among white voters without a college degree.
The Biden campaign certainly seems to think it’s doing well. Mr. Biden visited Iowa on Friday, and there’s not really any chance he’d be visiting there if the campaign didn’t think he was tied or ahead. Of course, campaigns have been wrong before. But so has Ms. Selzer.
No other indications of a last-minute surge. There’s another respect in which the Selzer poll is a bit of an outlier: It shows Mr. Trump gaining seven points compared with its prior survey. Four years ago, the Selzer poll was not alone in showing Mr. Trump picking up three points compared with a prior poll in early October, after the first debate.
This year, there’s not any other indication of a seven-point swing toward Mr. Trump, and there’s no reason to expect it either. The news environment doesn’t seem to have changed. The coronavirus spike is still the biggest national story, and I wouldn’t expect that to help the president. Maybe the final wave of polls over the next couple of days will hint at a shift, but for now there isn’t one.
CNN polls. A different batch of polls Saturday didn’t show much of a swing toward Mr. Trump: CNN/SSRS released high-quality polls in Wisconsin, North Carolina, Michigan and Arizona. Mr. Biden led in all of them, and usually by a comfortable margin.
The Arizona and Wisconsin results were pretty near the average, but the North Carolina result is worth an extra word. CNN/SSRS had Mr. Biden up by six points, the second day in a row a poll put him ahead by six points in the state.
North Carolina often gets overlooked on the battleground map. It doesn’t get the attention of Georgia or Texas, and it’s a relatively conservative battleground that hasn’t figured very prominently in Mr. Biden’s easiest paths to victory.
The polls this week at least raise the possibility that’s changing. Our average now puts Mr. Biden up by more in North Carolina than in Florida.
Pennsylvania. Don’t let Iowa make you lose sight of the most important state in the election: Pennsylvania. We got an additional data point there today from Muhlenberg College, which showed Mr. Biden with a five-point lead in the state. That’s slightly better for Mr. Trump than our average or its last poll of the state about a week ago, when Mr. Biden led by seven points. These sample sizes are small, so this kind of change could very easily just be noise. But I do think the majority of telephone polls in Pennsylvania over the last week or two have shown a subtle shift toward Mr. Trump. We’ll get a lot more from Pennsylvania on Sunday or Monday.
Looking ahead. Tomorrow at 5 a.m., you get the final Times/Siena polls of the cycle in Arizona, Florida, Pennsylvania and Wisconsin. I’d guess that all of the other major players will be weighing in with their final polls of the cycle over the next two days as well.
Oct. 31
A running table of all the presidential polls released today, with newest entries first:
| Pollster | Margin | Diff. from ’16 result | |
|---|---|---|---|
| U.S. |
U.S.C. Dornsife
7:34 AM ET
Oct. 18-31, 5,364 L.V.
|
Biden +12
54-42
|
+10D |
| Fla. |
St. Pete Polls
7:15 AM ET
Oct. 29-30, 2,758 L.V.
|
Biden +1
49-48
|
+2D |
| Mich. |
EPIC-MRA
7:15 AM ET
Oct. 25-28, 600 L.V.
|
Biden +7
48-41
|
+7D |
| Fla. |
RMG Research
7:09 AM ET
Oct. 28-30, 1,200 L.V.
|
Biden +4
51-47
|
+5D |
| Ariz. |
New York Times/
Siena College 5:54 AM ET
Oct. 26-30, 1,252 L.V.
|
Biden +6
49-43
|
+10D |
| Fla. |
New York Times/
Siena College 5:54 AM ET
Oct. 27-31, 1,451 L.V.
|
Biden +3
47-44
|
+4D |
| N.M. |
Research & Polling Inc.
5:54 AM ET
Oct. 23-29, 1,180 L.V.
|
Biden +12
54-42
|
+4D |
| Pa. |
New York Times/
Siena College 5:54 AM ET
Oct. 26-31, 1,862 L.V.
|
Biden +6
49-43
|
+7D |
| Wis. |
New York Times/
Siena College 5:54 AM ET
Oct. 26-30, 1,253 L.V.
|
Biden +11
52-41
|
+12D |
| Fla. |
ABC News/
The Washington Post 12:14 AM ET
Oct. 24-29, 824 L.V.
|
Trump +2
48-50
|
<1R |
| Pa. |
ABC News/
The Washington Post 12:14 AM ET
Oct. 24-29, 810 L.V.
|
Biden +7
51-44
|
+8D |
Advertisement
And keep an eye out for the Selzer poll Saturday.
We didn’t get very much polling today, with most pollsters saving their final polls for release over the weekend or on Monday. So it’s a fine time to take stock of what we know and still don’t know after today’s data, heading into the final round of polls.
State polls
| Pollster | Margin | Diff. from ’16 result | |
|---|---|---|---|
| Fla. |
YouGov
Oct. 16-26, 1,200 L.V.
|
Biden +2
48-46
|
+3D |
| Fla. |
P.P.P. (Dem. pollster)
Oct. 28-29, 941 voters
|
Biden +7
52-45
|
+8D |
| Ga. |
Landmark Communications
Oct. 28-28, 750 L.V.
|
Trump +1
47-48
|
+4D |
| Mich. |
RMG Research
Oct. 27-29, 800 L.V.
|
Biden +7
51-44
|
+7D |
| Mich. |
Kiaer Research
Oct. 21-28, 669 L.V.
|
Biden +13
54-41
|
+13D |
| Mich. |
P.P.P. (Dem. pollster)
Oct. 29-30, 745 voters
|
Biden +10
54-44
|
+10D |
| N.C. |
Rasmussen Reports/
Pulse Opinion Research Oct. 28-29, 800 L.V.
|
Trump +1
47-48
|
+3D |
| N.C. |
Cardinal Point Analytics (CardinalGPS)
Oct. 27-28, 750 L.V.
|
Trump +2
46-48
|
+2D |
| N.C. |
East Carolina University
Oct. 27-28, 1,103 L.V.
|
Biden +2
50-48
|
+6D |
| N.C. |
Marist College
Oct. 25-28, 800 L.V.
|
Biden +6
52-46
|
+10D |
| Pa. |
P.P.P. (Dem. pollster)
Oct. 28-29, 1,012 voters
|
Biden +7
52-45
|
+8D |
National polls
| U.S. |
Rasmussen Reports/
Pulse Opinion Research Oct. 27-29, 1,500 L.V.
|
Biden +3
49-46
|
<1D |
| U.S. |
Fox/Beacon/Shaw
Oct. 27-29, 1,246 L.V.
|
Biden +8
52-44
|
+6D |
| U.S. |
Opinium
Oct. 26-29, 1,252 L.V.
|
Biden +14
55-41
|
+12D |
| U.S. |
IBD/TIPP
Oct. 25-29, 959 L.V.
|
Biden +6
51-45
|
+4D |
| U.S. |
U.S.C. Dornsife
Oct. 16-29, 5,330 L.V.
|
Biden +12
54-42
|
+10D |
| U.S. |
Hofstra University
Oct. 19-26, 2,000 L.V.
|
Biden +11
54-43
|
+9D |
| U.S. |
Long Island University
Oct. 26-27, 1,573 adults
|
Biden +11
47-36
|
+9D |
What we know: Biden’s lead didn’t change. Joe Biden holds a nine-point lead nationwide, according to our average. It’s the largest lead a candidate has had at this stage since Bill Clinton in 1996.
There’s been no sign of tightening since the last debate, and I don’t see any sign of it in the few polls we got Friday either. A Fox News poll showed Mr. Biden up by eight points nationally, about the same as it has shown in past polls. Two of the president’s best surveys — IBD/TIPP and Rasmussen — actually ticked toward Mr. Biden.
Most of the state polling was also pretty decent for Mr. Biden, who appears to hold a very clear lead in the states carried by Hillary Clinton plus Michigan, Wisconsin and Nebraska’s Second Congressional District. It puts him just one more bigger-than-Iowa-sized Trump-held battleground state away from the presidency. The best candidate is Pennsylvania.
What we’ll learn over the final stretch: Pennsylvania. We have had relatively few polls of Pennsylvania this week, in part because most pollsters are saving the best for last (ourselves included). But most of the polls we have had of Pennsylvania would count as good but not quite great for Mr. Biden.
He’s up maybe six points on average, which is fine but in a tier beneath the other Northern battlegrounds, like Wisconsin, Michigan and New Hampshire. And the trend line hasn’t been great: Most of the recent polls of the state have shown President Trump is gaining, even if incrementally.
We should get more clarity on this question soon. NYT/Siena is polling there now, and I think we have every reason to expect that all the big players will weigh in one more time before the election, including ABC/Washington Post, Monmouth, CBS/YouGov and more.
I can’t repeat this often enough: Whether Mr. Biden has a significant lead or only a modest one really comes down to Pennsylvania. If he’s up by eight points in Pennsylvania, the president is looking for a miracle. But if it’s only four points, Mr. Trump’s in range where a fairly normal polling error might get him over the top — like four years ago.
What we’ll learn over the final stretch: Selzer in Iowa. On Saturday night, we’ll get the final Iowa poll from the vaunted Des Moines Register/Ann Selzer poll.
The survey will probably be the last, good look at the Upper Midwest by a top-tier pollster, and the results will help shape the way we think about the region and the state of the race over all. Four years ago, the final Selzer poll showed Mr. Trump jumping to a seven-point lead in Iowa over the final stretch — the clearest sign at the time of his late surge in the Midwest.
This year, Iowa is an interesting battleground state in its own right, reflecting Mr. Biden’s big gains among white voters. Recent polls show him ahead in the state, which ultimately voted for Mr. Trump by almost 10 points four years ago. If Selzer shows a close race or even a Biden lead, it’s very solid confirmation that the president’s at a big deficit across the Northern tier. On the other hand: Many Biden supporters will be quite unnerved if the Selzer poll again shows the president with surprising strength at the finish.
Oh, there’s a high-profile Senate race there, too.
What we’ll learn over the weekend: Did the president gain even a little bit? For as long as I’ve been writing this nightly polling diary, I’ve searched and searched for signs of “tightening” — after an event like a presidential debate, after good news for the president, or even after a spell of just not-bad news. It hasn’t really happened. But I will be looking for such movement as we get our final wave of national surveys.
No, there’s no obvious reason to expect tightening this time either. Nothing has really popped in the news, as far as I know. Despite tumultuous events, it has been a very stable polling year. But late tightening has happened before, even without an obvious cause. In fact, it happened in 2016: Mr. Trump started gaining in the polls released about 10 days after the final debate, before the Comey letter.
If Mr. Trump can get a few solid results at the end, showing him down more like six or seven points instead of nine points, it will nudge him a little closer to the range where he could realistically hope for things to break his way in the Electoral College.
Next: Times/Siena in Arizona, Florida, Pennsylvania and Wisconsin.
Tuesday: Election Day. More on that later.
Oct. 30
A running table of all the presidential polls released today, with newest entries first:
The race is steady, but here are the key numbers to look for.
Another deluge of polls from across the country today, from some of the country’s best-known pollsters and from some even I have never heard of.
State polls
National polls
| U.S. |
Rasmussen Reports/
Pulse Opinion Research Oct. 26-28, 1,500 L.V.
|
Biden +1
48-47
|
+1R |
| U.S. |
Morning Consult
Oct. 26-28, 16,374 L.V.
|
Biden +9
52-43
|
+7D |
| U.S. |
AtlasIntel
Oct. 26-28, 1,726 L.V.
|
Biden +5
51-46
|
+3D |
| U.S. |
Harris Insights & Analytics
Oct. 25-28, 2,359 L.V.
|
Biden +6
53-47
|
+4D |
| U.S. |
IBD/TIPP
Oct. 24-28, 945 L.V.
|
Biden +5
50-45
|
+3D |
| U.S. |
U.S.C. Dornsife
Oct. 15-28, 5,279 L.V.
|
Biden +12
54-42
|
+10D |
| U.S. |
Suffolk University
Oct. 23-27, 1,000 L.V.
|
Biden +7
51-44
|
+5D |
| U.S. |
YouGov
Oct. 20-27, 1,500 L.V.
|
Biden +9
53-44
|
+7D |
| U.S. |
Winston Group
Oct. 23-26, 1,000 R.V.
|
Biden +5
48-43
|
+3D |
| U.S. |
Cometrends
Oct. 13-26, 2,500 adults
|
Biden +12
56-44
|
+10D |
The big picture. Another day of polls, another day without good news for the president. Here’s one easy way to tell: There’s a lot of blue on these tables. In particular, there’s a lot more blue in the right-most column, which represents the difference between a poll result and the 2016 election result.
As was the case yesterday, I’m not sure there’s a single poll on Thursday that qualifies as good news for the president. A lot of the red came from one firm: Rasmussen Reports. We’ve talked about Rasmussen’s tendencies before. We haven’t talked about a lot of the other national polls showing a mid-single-digit race. We’ll get to them in a bit.
What else might count as good news for the president? Maybe being ahead by four points in Texas, as RMG Research found? Then again, it probably isn’t a good day if your best news is a lead in the state that represents your 163rd electoral vote.
Florida. Today’s big focus is Florida, where we received three live-interview surveys showing Joe Biden with a modest lead: Monmouth showed him up by five points (averaging its two likely-voter scenarios), while Marist and Quinnipiac showed Mr. Biden up by four and three points.
If Mr. Biden wins Florida, the race is all but over. So a clear Biden lead here, as these polls suggest, is a big deal.
That said, this particular set of polls might not be quite as great for Mr. Biden as it looks. Our poll average is largely unconvinced: Mr. Biden still leads by only two points.
One factor is that Quinnipiac and Monmouth have both tended to have favorable results for Mr. Biden in Florida. It doesn’t mean they’re wrong, but it does mean that these polls don’t do as much to revise our thinking as you might expect. The new Quinnipiac poll is eight points better for President Trump than last time!
Of the three results, the Marist poll is the only one that shows movement toward Mr. Biden, with a four-point swing in his favor. But this is a poll that doesn’t ensure it has the right number of less educated voters, which often means it understates the president’s standing.
Why Florida is hard. Even if Mr. Biden did hold a more comfortable lead in Florida, I don’t think you’d find Biden campaign staffers sleeping easy. Florida is a tough state for pollsters, in no small part because of its large Hispanic vote.
In most states — as well as in most of Florida — the Hispanic voters easiest to reach might be suburban, affluent, native English speakers who tend to tilt Republican compared with more Democratic, working-class, urban, Spanish-speaking Latinos. That’s always tough.
But in Florida, there’s another twist. The very hardest group of voters to reach is a staunchly Republican group: Cuban Hispanics in Miami-Dade County, whose views are so different from the rest that it creates a big danger for pollsters. Giving more weight to Hispanic voters from the rest of the state might do the exact opposite of what you want: You’d be replacing Republican Cuban voters in Hialeah with Democratic Puerto Ricans in Kissimmee.
Worst of all, there’s really not much you can do about it as a pollster. Cuban voters are around 4 percent of the statewide vote. So you’re hoping to get maybe 40 of them in a statewide poll of 1,000 people? You just can’t create a design that addresses such a small group, at least not well. Even if you did it well, nailing such a tiny subgroup is a matter of luck as much as anything.
Not surprisingly, these three pollsters have some very different takes on the state’s Hispanic vote. NBC/Marist found Mr. Trump ahead among Florida Latinos, 52-46. On the other end, Monmouth found Mr. Biden up, 58-32.
Nationally. At first glance, there are a fair number of national polls that look decent for the president. There are a lot fours, fives and sixes on our table, in contrast with Mr. Biden’s nine-point lead in our average.
On closer inspection, these results aren’t so meaningful. For one thing, most of these firms are of uncertain quality. More important: Most show no change from their prior surveys, suggesting the president’s strong result reflects their methodological choices, not a shift.
The last Harris poll was Biden plus-four, and now he’s up six; the last AtlasIntel poll showed Biden up three; now he’s up five. Winston Group showed Biden plus-five, but before it was plus-four. IBD/TIPP shows Biden up five, about where it has stood over the last week on average. Rasmussen is Rasmussen. And so on.
The more frequent and familiar pollsters — YouGov, U.S.C., Morning Consult — are all more or less showing an unchanged race, and that’s bad news for Mr. Trump
Pennsylvania? We’ve gotten a pretty complete picture of the battlegrounds over the last few days, with a lot of polls in Florida, Michigan, Georgia, Wisconsin and North Carolina — including a new Times/Siena poll today showing Mr. Biden up three points in North Carolina. One place we’ve seen less of is Pennsylvania.
Late yesterday, a Franklin & Marshall poll showed Mr. Biden up six points in the state — unchanged from September, though down from the summer. Quinnipiac showed Mr. Biden up seven points, but that’s also down from its prior polling, which showed Mr. Biden up by eight, 13 and eight points over the last few months. A few other polls showed a race in the mid-single digits.
If you were asked to find a soft spot in Mr. Biden’s armor in the post-debate polling, I think Pennsylvania would probably be the place to look. The results aren’t quite as good for him as those in similar states, like Michigan and Wisconsin or even New Hampshire — where today three polls showed Mr. Biden up by double digits.
I don’t want to overstate the weakness for Mr. Biden here: He’s up six points on average. But I can’t overstate the significance of whether Mr. Biden’s way ahead or only modestly ahead in Pennsylvania.
For as long as this blog has been around, we’ve noted that Mr. Biden seemed to hold a clear lead in the Clinton states plus Michigan and Wisconsin — leaving him just one decent-size state short of the presidency. The problem for Mr. Biden: He hasn’t always held a big lead in another state. Pennsylvania is Mr. Biden’s best option. He has occasionally held a big lead there; he’s never really had a big lead in, say, Florida or North Carolina.
As a result, it makes a big, big difference whether Mr. Biden’s up by, say, four or eight points in the state. If it’s four or five, we might go into election night thinking Mr. Trump has an outside but very real chance to win the state — and therefore the election. If the polling lead is seven or eight, it would take a polling miss bigger than the one in Wisconsin in 2016. That’s still possible, but it just doesn’t happen that often.
Today, our average is right in the middle of those two possibilities: Biden by six points. And with relatively few recent polls recently, it’s not hard to imagine that the next wave of surveys in the state might tilt us one way or another. If they do, it will bring a lot of clarity to the state of the race.
Next: NYT/Siena final wave, including in Pennsylvania. More on that soon.
Advertisement
Biden holds small but durable lead; Democrat has narrow edge in Senate contest.
My colleagues Trip Gabriel and Isabella Grullón Paz have the story.
New York Times/Siena College poll of likely voters in North Carolina
Oct. 29
A running table of all the presidential polls released today, with newest entries first:
Not a good day for Trump.
For a couple of days now, I’ve been telling you to wait — that the polls are coming. Today they came. And they told a very clear story.
The big picture. There were a lot of polls Wednesday, and I could go on about this poll or that one. But let’s frame it this way: Was there a single state poll that qualified as good news for the president? Maybe the Ipsos poll showing him down only a point in Arizona? Otherwise, I don’t think so.
In the right-hand column of our table, only one state poll showed him within three points of his election result in a state in 2016 (it’s blue numbers all the way down). I don’t see anything suggesting any tightening, and a lot of today’s polls were conducted entirely after the final debate.
There’s just no good news for the president here, unless you want to count the Rasmussen poll — which we’ve talked about before.
State polls
National polls
| U.S. |
J.L. Partners
Oct. 26-28, 844 L.V.
|
Biden +14
55-41
|
+12D |
| U.S. |
YouGov
Oct. 25-27, 1,365 L.V.
|
Biden +11
54-43
|
+9D |
| U.S. |
Rasmussen Reports/
Pulse Opinion Research Oct. 25-27, 1,500 L.V.
|
Trump +1
47-48
|
+3R |
| U.S. |
IBD/TIPP
Oct. 23-27, 991 L.V.
|
Biden +5
50-45
|
+3D |
| U.S. |
Ipsos
Oct. 23-27, 825 L.V.
|
Biden +10
52-42
|
+8D |
| U.S. |
Global Marketing Research Services
Oct. 23-27, 1,006 L.V.
|
Biden +14
53-39
|
+12D |
| U.S. |
U.S.C. Dornsife
Oct. 14-27, 5,269 L.V.
|
Biden +12
54-42
|
+10D |
| U.S. |
Redfield & Wilton Strategies
Oct. 25-26, 4,790 L.V.
|
Biden +10
51-41
|
+8D |
| U.S. |
SSRS
Oct. 23-26, 886 L.V.
|
Biden +12
54-42
|
+10D |
| U.S. |
Qriously
Oct. 22-26, 2,234 L.V.
|
Biden +10
49-39
|
+8D |
| U.S. |
Change Research
Oct. 23-24, 1,125 L.V.
|
Biden +8
51-43
|
+6D |
| U.S. |
Spry Strategies
Oct. 20-23, 3,500 L.V.
|
Biden +2
48-46
|
<1R |
| U.S. |
Rethink Priorities
Oct. 20-20, 4,933 L.V.
|
Biden +9
51-42
|
+7D |
| U.S. |
GBAO (Dem. pollster)
Oct. 15-19, 1,150 R.V.
|
Biden +13
53-40
|
+11D |
What is going on in Wisconsin? If anything had poll junkies talking, it was the two big and conflicting results in Wisconsin. ABC News/Washington Post, a high-quality pollster, found Joe Biden leading by … an eyebrow-raising 17 points. (Not a typo — or not a joke, as Mr. Biden might say.) Later in the day, Marquette University Law School, perhaps the most trusted pollster in the state, found him leading by five points. You probably expect me to dive in and explain the difference. Unfortunately, I don’t have answers.
But I can explain why wild results are relatively likely in Wisconsin. It’s a hard state to poll. There’s very little to help pollsters make sure they get the right number of Democrats and Republicans. There’s no party registration. There isn’t a large racial divide, so it’s not the kind of place where making sure you have the right number of white and Black voters is tantamount to having the right number of Ds and Rs. And so on.
But I can’t explain why these two results are so different. These two pollsters are actually quite similar. They use the same basic methods, and they polled over nearly the exact same field period. To me, this suggests that it’s just statistical noise: the kind of weird, outlying results you inevitably get with random sampling, especially when you can’t count on partisan weighting or race to get you back into a reasonable spot.
You might also expect me to say one poll result is right and the other is wrong. But it’s not that simple.
Take the ABC/Post poll. Yeah, it’s a real outlier. Mr. Biden entered the day with a nine-point lead in Wisconsin according to our poll average, and the ABC/Post result is the largest lead in the state by any public pollster this cycle. So no, I don’t think Mr. Biden is “really” up 17 points in the state.
But that doesn’t mean it should be dismissed out of hand. After all, would you have dismissed an ABC/Post poll that showed Mr. Biden up just one point, which represents the same difference from the average but in the other direction? That would have rightly moved your view of the race toward the president, and this ought to move your view toward Mr. Biden.
The composition of the electorate — Democrats outnumbering Republicans by four points — is a little bit more Democratic than I would have guessed, but it’s no reason to throw the poll out. It’s also not that different from the Marquette poll, which had the two parties evenly split.
If there’s one red flag in the ABC/Post poll, it might be that it found more registered voters who say they backed Hillary Clinton in 2016 than who say they backed President Trump. The survey would have found Mr. Biden up by 12 points if it had adjusted the sample to be Trump plus-one in 2016, like the actual result. If you want to make that adjustment in your head, you’re totally welcome to do so.
But it’s not necessarily the best practice. Voters can misrepresent how they voted, especially if they’ve soured on the person they once backed. And polls don’t represent the 2016 electorate; they represent the 2020 electorate. People have died and moved since 2016, and it’s not at all clear how that’s moved the needle. The fact that pollsters are representing 2020, not 2016, creates another problem: You didn’t design your survey to represent the 2016 vote, and so you shouldn’t expect it to mirror it exactly. For instance, say you have the right number of Democrats and Republicans in your poll, but you have too many nonvoting Democrats and too many voting Republicans. If so, your unbiased poll would have too many Trump supporters.
If you really wanted to be critical, you could also find issues with aspects of the Marquette poll. You could say their young voters or their sample of the city of Milwaukee were way too Republican.
So what’s the upshot? It’s quite fair to say Mr. Biden has the lead in Wisconsin. It’s an uncertain one, to be sure. Maybe he has opened up a double-digit lead, as coronavirus cases mount in one of the nation’s biggest hot zones. If Biden did win by a huge margin, I have no doubt folks would feel comfortable attributing it to that. Or maybe it’ll be a lot closer to the Marquette result. Regardless, there’s no way to avoid concluding that Biden has a solid advantage.
Here’s a bit of trivia: If Mr. Trump won Wisconsin in the end, this ABC/Post poll would be only the second poll of a regularly scheduled statewide general election in the big FiveThirtyEight database of polls since 1998 to show a winning candidate behind by so much in the final three weeks.
The record-setter is a Rasmussen Reports poll in the Michigan Senate race in 2000. It found the Republican incumbent senator, Spencer Abraham, leading the Democrat Debbie Stabenow, 53 percent to 34 percent, with 20 days to go. Ms. Stabenow won and has been a senator ever since.
The rest of the North. If you want a little more clarity on Wisconsin, one option is to consider demographically similar states. Today, we got two polls showing Mr. Biden up by eight and seven in Michigan, from NYT/Siena and ABC/Post. There was a Colby College poll showing Mr. Biden up double digits in Maine, which is another mostly white Northern tier state full of Obama-Trump voters. There was even a Gravis poll yesterday showing Mr. Biden up double digits in Minnesota.
Put it together, and it would be no surprise if Mr. Biden’s up comfortably in Wisconsin. But there’s even less reason to believe he’s up by 17.
A shift toward Biden in Georgia. After the Wisconsin polls, the most important result came out of Georgia, where Monmouth had Mr. Biden up by two to five points, depending on the turnout scenario. This is a big shift toward him from its last poll, which showed Mr. Trump ahead by between one and five points.
I don’t know if I think Mr. Biden is really up by as many as five points in Georgia, but I do think there’s plenty of evidence that the state has been shifting toward Democrats. There has been a lot of reporting in the last week that things are breaking Democrats’ way in Georgia, and Mr. Biden even decided to visit the state.
Nationally. It’s a somewhat odd set of national polls, but the major pollsters all had positive news for Mr. Biden. YouGov/Economist had him up by 11 points, ticking up from its pre-debate poll. CNN/SSRS had him up by 12, a fine result for him even if it’s a pollster that has put up some particularly strong results for him before. Otherwise, just look at the average. A solid Biden lead.
Tomorrow. NYT/Siena in North Carolina, our final poll of the state.
Then! Arizona, Florida, Pennsylvania and Wisconsin to finish out the campaign. We’re so close.
Advertisement
Biden holds an eight-point lead in our final poll of the state.
My colleagues Jeremy W. Peters and Isabella Grullón Paz have the story.
New York Times/Siena College poll of likely voters in Michigan
Posts
No comments:
Post a Comment