Zero Days Out: Final Predictions

I did it! After a long layoff, PPFA has delivered on its promise of one post per day for the last week of the election. I’m not sure whether to say “You’re welcome” or “Sorry,” so I’ll just say: I’m so grateful to have you here.

But enough about me. Let’s talk about you. Let’s talk about the country. IT’S ELECTION DAY! You made it. The country has forged through a pandemic, social strife, and a Lakers NBA Championship just to get to this election, a process we should never take for granted. In the history of civilization, republican institutions are not the rule. They are the exception. In so many ways we lament our modern misfortunes, but in so many more ways we should remind ourselves how lucky we are to be alive right now.

Enough mushy stuff. You didn’t come here for mush. You came here for rock hard PPFA analysis. It’s now time to lock in predictions for Election Night 2020, even if we might not be able to check my work for days. This post will be split into three parts, the last of which will be the longest.

1. Part I: the House of Representatives
2. Part II: the Senate
3. Part III: the Presidency

As you watch tonight, remember to keep handy yesterday’s post, where I walked you through poll closing times and the states to keep an eye on.

Okay, let’s do it — PPFA’s official 2020 prediction begins… NOW!

Part I: The House of Representatives

One week ago, back when you were at peace without giving 15 minutes of your precious time to this wordy website, I started my daily posts with an assertion that of the three pieces of government up for grabs tonight, it was the House that was easiest to predict. I’ve seen nothing in the last week to back me off of that claim.

Generic Ballot: Democrats +6.8 (one week ago: Democrats +6.6)

The Cook Report’s Report:

House Prediction: Tonight, the Democratic Party will retain the House. Putting a specific number is a fool’s mission, but I’ll guess that the Dems tick up a few seats to get to about 240 while the GOP clocks in around 195.

Part II: The Senate

I took two days to examine the super close race for the Senate. I left thinking the Democrats would get to 50, the Republicans to 48, and neither Georgia Senate race would see its winner get to 50%, which triggers run-off contests on January 5. I noted, however, that I wanted to see more data on a few races, and we did get a bit more since Saturday.

None of that data changed my mind, however. I’ll lock in what I had before:

• Democrats: 35 seats not up for re-election + 10 safe seats: 45
  • Plus MI, CO, AZ (some confidence) for 48
  • Plus ME and barely NC (low confidence) for 50
• Republicans: 30 seats not up for re-election + 9 safe seats: 39
  • Plus AL, MS, KY, TX, AK, KS, and SC (some confidence) for 46
  • Plus MT and IA (low confidence) for 48
• January 5 runoffs: both Georgia elections

Part III: The Presidency

And now the big kahuna. Yesterday’s post left you with the broadest possible battleground map:

We can pencil in Biden with a 212 – 125 lead with 13 states and two Congressional districts (Maine and Nebraska allow their Congressional districts an electoral vote, which is getting really annoying to repeat) for a total of 15 contests to figure out.

Before I get into individual contest predictions, here’s a big-picture chart, which lists FiveThirtyEight’s polling averages of all the swing states:

If we were to say there’s an across-the-board 3-point polling error to President Trump, we would give him all states below Pennsylvania there. Even with that sweep, Trump would still be short of 270 electoral votes (258). It will require him to sweep those states and then win Pennsylvania’s 20 electoral votes, or one to three states in which polling is even worse for him. And again, that’s if he sweeps all the close states.

Now let’s say the polling error goes in the other direction. Let’s say Biden wins all the state he leads and the states where Trump is up a couple points. The final result is a whooping 413 to 125.

In other words, a 3-point polling error in either direction can make a huge difference. It’s not easy being a presidential prognosticator, I’ll tell you that.

I can’t wait any longer. It’s time to start whittling and awarding. Again, we’re starting at 212-125. I’ll progress from highest to lowest confidence:

Tier 1: High confidence

New Hampshire — 4 Electoral Votes (EVs): my highest confidence returns us to the Granite State, where we once spent months analyzing the Democratic Primary. It was once was a swingier state that voted for George W. Bush in between the Clintons and Obama, and it’s usually close. However, since 2000, it has been, like an old horror movie, slowly enveloped by the blob that is Democratic New England. Biden has led the last ten polls by 7+ points, half of them by double digits. Biden 216, Trump 125

Texas — 38 EVs: I’ve foreshadowed this pick for a week. Democrats are going to get their hearts broken in Texas. Again. A Democrat hasn’t won a statewide race there since 1994, a period of over a hundred contests. Though a Dallas Morning News poll found Biden up 3 points a couple weeks ago, polling since then suggests Trump should win a close one. I wouldn’t bet the other way until it happens. Biden 216, Trump 163

Michigan — 16 EVs: Michigan was a huge surprise in 2016. Hillary Clinton’s average lead in the polls was 3.4 and Trump won by 0.3, a difference of 3.7. This year, Biden’s average RCP lead is 4.2 (and no pollster other than openly right-leaning Trafalgar and Insider Advantage has found the race closer than 7 in the last month), so an identical polling error would still result in a Biden victory. Biden 232, Trump 163

Tier 2: Some confidence

Minnesota — 10 EVs: Biden’s lead in Minnesota (+4.3) isn’t as robust as his lead in Michigan, but there are reasons for Democrats to feel comfortable. First is that Minnesota voted for Clinton in 2016. Though it was her narrowest win (by 1.5%), very little about the current political climate suggests Trump is more popular now than he was four years ago. Second, even Trafalgar has Biden up 3 in the state. I can’t imagine it has undercounted Republican support. Biden 242, Trump 163

Iowa — 6 EVs: I was already thinking this was a Trump win a week ago, and that was before the highly respected Selzer poll came out of Iowa showing Trump up 7. He’s led the last four polls out of the state. With momentum, he should defend it successfully. Biden 242, Trump 169

Tier 3: Low confidence

Nevada — 6 EVs: Like Minnesota, it’s a state carried by Hillary Clinton. The main mitigating circumstance here is that Biden is less popular with Hispanic voters than was Clinton, and Nevada is a heavily Hispanic state. Though three polls in the last month showed Nevada as a six-point race, another three show it as a two-point race, and Trafalgar predictably slid in at the end with a one-point lead for the President. Still, if Trafalgar is showing Trump with just a one-point lead, it’s probably a Biden state like all the other polls suggest. Biden 248, Trump 169

Wisconsin — 10 EVs: Biden’s average polling lead in Wisconsin (+6.7) is stronger than in many of the states above. However, it more accurately mirrors Trump’s strongest demographic: white voters without a college degree. Wisconsin, at 83% white, tops all Rust Bell states in white population, and the only Rust Belt state to compete with Wisconsin in that category, Minnesota, has a solidly higher portion of college-educated voters. In short, if pollsters are undercounting Trump voters, it’s likely in Wisconsin where that’s writ large. The 2016 polling error in Wisconsin was enormous; Clinton had an average polling lead of 6.5, but Trump won by 0.7, a swing of 7.2%. That’s nearly twice the error in Michigan and almost thrice the error in Pennsylvania.

That said, Biden’s lead in Wisconsin looks as safe as one can expect. For every poll that has Trump within 3 (including Trafalgar’s one-point spread), there’s a poll that has him up double digits. Biden 258, Trump 169

---

Let’s pause for a quick map update here before I throw darts at my Tier 4. Remaining are the final eight contests, which includes six states and two districts:

Now, of course I can be wrong in the allotted states so far, but I’d like to think that so far I had good reasons for each allotted state. If those predictions hold, we can see why Biden has a great shot to win. Just 12 short of a 270 majority, nearly any single state or Arizona plus a Congressional district gets him there. Trump, meanwhile, has to win just about everywhere.

It’s an uphill but not at all impossible climb, particularly, as referenced before, if there’s a polling error in each of the following very close states. Let’s see if Biden can find 12 electoral votes to emerge victorious.

Tier 4: No confidence (alphabetical order)

Arizona — 11 EVs:

As will be the theme for most of these states, it’s close in Arizona. Trump has closed what once was a sizable gap between the two candidates, momentarily taking the lead in the Real Clear Politics average last week, but Biden has nosed in front again. The two Republican-leaning polls like Trump, but of the seven polls that make up the average, it is the NYT/Siena poll I trust most. It’s a high quality pollster, it has the largest sample size, and it shows Biden up 6. Meanwhile and unusually, Arizona was one of the few states that overestimated Trump’s support in 2016. We’re seeing Democrat Mark Kelly poll well in the state’s important Senate race, so I think we’re seeing a blue state this time around. Biden 269, Trump 169

It looks like Biden gets to 269 here, which clinches at least a tie. However, as I wrote last week, I think a tie goes to the Republican these days, so he really needs to find 270.

Florida — 29 EVs:

Here’s another state where the polls are tight and divided between conservative pollsters and the herd. Beyond that, I think early ballots tell a pretty convincing story. Unlike most states, where registered Democrats turned in ballots at a far higher rate than their Republican counterparts, in Florida it’s much closer, with momentum to Republicans. As for day-of voters, the GOP is certainly at the advantage.

Ultimately, I like Trump’s momentum here more than in Arizona. Also keep in mind that despite the 2018 Democratic wave, the GOP won the governor’s mansion in that election. Florida may be insulated from political movements in the rest of the country. Biden 269, Trump 198

Georgia — 16 EVs:

Georgia has come home to Trump, at least according to the polls. I think we have to get the edge to the President here. Not only has he led the four most recent polls, but he won the state by over 5 points in 2016. I just don’t think he’s 5 points less popular in what has been a solidly red state since Bill Clinton’s first election. Biden 269, Trump 214 (Uh oh, Biden’s having trouble sealing the deal!)

Maine’s Second Congressional District — 1 EV:

Here’s every poll from Maine’s second Congressional District, or, as I like to call it, #Me2. We can see polling is too scant for an RCP average, perhaps because this extremely rural region is just now getting telephones. Of course, this rural nature is why the district votes so Republican lately:

In 2016, Trump won the district by 12 points, the greatest spread of all today’s listed contests. I bet he wins it again thanks to some undercounted Trump supporters. Chalk another win up for the President. Biden 269, Trump 215 (Trump is lurking!)

Nebraska’s Second Congressional District — 1 EV: We don’t have much polling here, but the polls we do have show Biden up. Plus, Nebraska’s CD2 is based around Omaha, already Nebraska’s most Democratic area, and that was before the Trump Campaign literally left people out in the cold, which feels like a metaphor for something. Biden 270, Trump 215

There it is! That’s 270. But since a 270-268 win would be tenuous, particularly when the margin came from Nebraska, Democrats would sure like some padding.

Three states to go.

North Carolina — 15 EVs:

Like Arizona and Florida, we see Trump closing strongly in another state he won in 2016. Just like in those states, I like that kind of momentum for him. Four years ago, when he surprisingly won the upper Midwest states, Hillary Clinton had the average polling lead but he was closing the gap across the region. Polling didn’t capture that final flood of Ken Bones in his direction. I swore to myself I’d favor late momentum over close polling this time around (although early voting and a much lower number of undecided voters are mitigating factors).

Trump’s comeback here looks complete, particularly in another state that underestimated Trump’s support last time around (by 2.7). Biden 270, Trump 230

Ohio — 18 EVs:

Although ostensibly we see Trump making a big push in Ohio, we should take caution. This late super surge relies on our friends at Rasmussen and Trafalgar giving Trump his largest Ohio leads of the year. (Fox News also has Trump up, but I actually think it does a great job polling. I’m skeptical of Ras and Traf.) Nonetheless, this state, by far the most Republican of the Rust Belt contests, was always fertile ground for Trump. The Quinnipiac and Emerson poll showed Biden’s lead dwindling even before the conservative pollsters showed those Trump leads. Biden likely built up a decent early voting lead before Trump’s late rise, but New York Times/Siena’s last poll there, from a few weeks ago, had Biden only up one. Biden 270, Trump 248

Whoa, Trump is one Pennsylvania away from giving us a 270-268 final result, which feels very litigationy to me. If I have every prediction right except one, and that one misstep is a PPFA-projected Biden win, then Pennsylvania will decide the election.

Deep breath

Pennsylvania — 20 EVs:

Democrats, if you have a paper bag handy, this is when you start breathing into it.

The “tipping point state” is now down to an RCP average polling gap of 1.2 points — and the polls in 2016 underestimated Trump by 2.6. Trump appears to have the momentum here, and remember what I swore myself to not overlook again — late momentum!

But I must say, the conservative pollsters were all hands on deck here to keep Trump’s voters motivated. Trafalgar I discussed last week. InsiderAdvantage is sponsored by the pro-Trump Center for American Greatness. Susquehanna Polling and Research specializes in representing Republican and other conservative groups.

Now, they may be right! (I may be crazy.) Trafalgar did great four years ago. Rasmussen too. Whether they were lucky or good might be decided today. Regardless, the fact remains that all three of them have Trump leads, which I think is screwing with the average.

Instead of over-relying on the RCP average in this pivotal state, I’d like us to use a wider-angle lens of all the recent polling:

Only the right-leaning pollsters ever have Trump tied or in the lead, and their last-minute flurry of polling bumped out pretty darn recent polls from the RCP average. Indeed, there’s some buzz that RCP hasn’t deployed an even strategy when determining what polls to cut off from the current averages.

Still, even Rasmussen said it was a 3-point spread to Biden a couple weeks ago, and they said it again this weekend. The large-sampled NYT/Siena poll has Biden up 6, and a majority of recent pollsters are within two points of that. Throw in the locked-in early vote, the lack of undecideds that are making up their minds late, and the President dismissing a virus that’s still killing 800 people a day as a media exaggeration, I don’t think Trump can make up this polling gap.

Pennsylvania is why the Democrats nominated Joe Biden. I think he wins it.

Biden 290, Trump 248

Release breath

Thanks for reading.

-PPFA