We Have a Lot of New Polls, but There's Little Sign of the Presidential Race Tightening

By Nate Silver, FiveThirtyEight

29 October 20

Instead, Biden gains in the Midwest.

fter a surprisingly sluggish weekend for polling, the floodgates have opened, with a mix of high-quality polls, low-quality polls and pretty much everything in between. And although there are some outliers in both directions, they tell a fairly consistent story, overall: A steady race nationally, perhaps with some gains for Joe Biden in the Midwest.

Let’s dive right in. Following our rules from my colleague Geoffrey Skelley’s post earlier this week, we’re going to look at every poll where at least half of the survey took place after last week’s debate, with a comparison to the most recent pre-debate survey by the same firm. If the pollster hasn’t surveyed a state before, we’ll make the comparison based on what the FiveThirtyEight polling average said in that state as of the night of the debate (Oct. 22) instead.

As an aside, I set a cutoff for polls as of 7 p.m. Eastern on Wednesday. (More will probably have filtered into the averages by the time that you’re reading this.) We’ll rotate through the four major regions of the country — the South, the West, the Midwest and the Northeast. But first, here are the post-debate national polls:

On average, Biden leads by 9 percentage points using a simple average of post-debate national polls, which matches his 9-point lead in our fancy-schmancy official FiveThirtyEight national average.

It’s a bit of a weird mix of polls, though, with more quantity than quality. The only two fully live-caller national polls in here are from CNN and CNBC/Hart Research/Public Opinion Strategies, which show Biden leading by 12 points and 11 points, respectively. On the one hand, the trendlines aren’t so bad for Trump in those live-caller polls. He trailed by 16 points in CNN’s previous national poll and by 9 points in the past Hart Research/Public Opinion Strategies poll (or by 11 if you prefer to compare it to the prior NBC/Wall Street Journal national poll, which uses the same polling team).

On the other hand, the higher-quality national polls are showing worse results for Trump than the lower-quality polls, which is rarely a good sign. More highly rated online firms — such as YouGov, Morning Consult and Ipsos — show Trump trailing by margins ranging from 9 to 12 points.

At least some of the fluctuations in our national polling average seem to reflect shifts in the ratio of high-quality and low-quality surveys. This can go through cycles, and recently the higher-quality firms have mostly been concentrating on state polling. Most of them will release their final national polls soon, though, and it wouldn’t be surprising if, following CNN and CNBC’s example, a few more of them show double-digit leads for Biden.

Something else to note: Although there’s been a slight decline in Biden’s national polls since the debate, a majority of state polls show his position improving. To be more precise, Biden has gained 0.7 percentage points in the average state poll since the debate, while he’s lost 0.5 points in the average national poll. That brings the national and state polls into better alignment after a period where national polls suggested that Biden led by 10 to 11 points but state polls were more consistent with a lead of about 9 points instead.

Let’s start our review of the regions with the South, where there’s been a lot of polling over the past couple of days:

It’s hard to tell a coherent story here because every state in the region seems to be marching to its own fight song. Biden’s gotten some great numbers in Georgia since the debate, but he’s had some terrible ones in Florida. North Carolina is somewhere in between, with polls variously suggesting that he’s gaining or losing ground.

Do I buy that there are these microfluctations between neighboring states? I’m not sure that I do, particularly since you have a different mix of pollsters from state to state. Florida and Georgia could use one or two more high-quality polls.

For the time being, though, Biden’s small lead in our Georgia polling average (+1.4) nearly equals his lead in Florida (also +1.4) and North Carolina (+2.1). That means Georgia needs to be taken seriously as a plausible — if unlikely — tipping-point state. It’s going to be the focus of a lot of resource expenditures over the final few days of the campaign, especially since it also has two competitive Senate races. Texas is a step or two away from the rest of the Southern group, meanwhile, as Trump leads in our polling average by 1.8 points there.

Next up on our whirlwind tour of America … the West:

This is a decent set of polls for Biden, including in Arizona, where Biden’s lead has oscillated between 3 and 5 points in our polling average for most of the year. It’s in the lower-to-mid part of that range now, at 3.5 points in our official polling average, although the raw average of post-debate Arizona polls is slightly better for him at +4.2.

I realize this might seem like splitting hairs, but Arizona is an important state for Biden, insofar as it’s probably his best backup option if he loses Pennsylvania. (If Biden loses Pennsylvania but wins Arizona, Michigan and Wisconsin, he’d also need to find one more electoral vote somewhere to break a 269-269 Electoral College tie, but he’d probably get that in Nebraska’s 2nd Congressional District, where he’s favored.)

In a world where Arizona is polling at Biden +5, it’s really almost as though he has a Plan A1 (Pennsylvania) and Plan A2 (Arizona plus NE-2). That would make his map more robust. In a world where Arizona is Biden +2 or Biden +3, conversely, it doesn’t really stand out from his other backup options, such as North Carolina and Florida, so it would be as though Biden has one Plan A (Pennsylvania) and a bunch of Plan Cs.

Now to the Midwest:

FiveThirtyEight’s Nathaniel Rakich covered this region this morning, so I’ll be brief here. This is the one part of the country where you can’t really say the race is steady; instead, this has been a strong set of polling for Biden, and he’s gained an average of 1.7 points in surveys in this region versus the pre-debate polls.

No, Trump probably doesn’t trail in Wisconsin by 17 points, as he does in the ABC News/Washington Post poll. (Good pollsters publish their outliers instead of sitting on them, and sometimes they turn out not to be outliers at all.) But average the ABC News poll with the three other post-debate polls of Wisconsin, and Biden is still ahead by 10.5 points there.

Biden’s polling gains may reflect that the COVID-19 outbreak in the Upper Midwest is bad right now, especially in Wisconsin. There is some evidence that the polls tend to shift against Trump following spikes in COVID-19 cases in a state, potentially a foreboding indicator for Trump as COVID-19 cases are also now rising sharply in many other states.

Finally, to the Northeast:

By the Northeast, as you can see from the chart above, I really just mean … Pennsylvania, plus one stray poll of Maine. This region has been underpolled. There haven’t been any live-caller polls of Pennsylvania since the debate, and the online and IVR polls show highly disparate results there. New Hampshire has had no post-debate polling at all, meanwhile.

We go back and forth here at FiveThirtyEight about whether Pennsylvania counts as part of the Northeast or the Midwest. I’ve usually been a proponent of the theory that for political purposes, Pennsylvania behaves like a Midwestern state. There are some important geographic, demographic and cultural differences between Pennsylvania and the Upper Midwest, however, and this year, the polling in Pennsylvania hasn’t tracked the numbers in Wisconsin and Michigan that well. Pennsylvania is also not currently experiencing as bad of a COVID-19 outbreak as those other states. So it may be less correlated with them than usual.

But one way or another, the final round of high-quality, live-caller polls in Pennsylvania is likely to make a big difference to our model. Pennsylvania is by far the most likely tipping-point state. If Biden gets up to, say, a 6- or 7-point lead there, he’ll be in a much safer position overall in the Electoral College than if he gets knocked down to a 4-point lead here instead. More polling of the Keystone State is surely to come soon.