Biden needs to do more than make up his deficit in national polls

If the popular vote is close, Trump will probably win the election.

Nate here, briefly! I’m excited since this is Eli McKown-Dawson’s first article for Silver Bulletin. And we’re making it a Model Talk (our weekly column for paid subscribers) since it goes into a lot of detail on the impact of the Electoral College on this year’s presidential forecast. Obviously, Democrats face some other big challenges — is Biden still going to be their nominee? But the Electoral College is a key background factor as they make their decision and will likely make their path more uphill whomever they pick. Bringing the race back a popular vote “tie” probably isn’t enough — they need some margin to spare to make up for Trump’s likely Electoral College edge. -NS

---

After last month’s debate, Donald Trump leads by about 3 points in our national polling average. But even now, the natural variation in polls means that we’re going to get individual polls that show a tied or even Biden +1 national race — just as we get others that show Trump ahead by 6 or 7. It’s obviously wrong to cherry-pick these polls to claim the race is tied. But things wouldn’t look great for Biden even if he and Trump were tied in the polling average. Why? Trump still has an edge in the Electoral College.

So for my first post that isn’t just updating the forecast landing page — remember when that was supposed to be only once a week? — I’m going to ignore all the drama about Biden and instead talk about Electoral College bias and why it should inform how we interpret polls like this one:

There’s nothing wrong with this tweet¹ — and honestly, being tied with Trump was relatively good news for Biden after a rough (to put it mildly) first debate. But the more general point here is that a tied national popular vote still isn’t good enough and would probably result in Trump winning the presidency.

Electoral College bias is smaller than in 2020 but still favors Trump

Simply put, we expect Trump to have an advantage in the Electoral College because the states most likely to decide the election, like Pennsylvania, are forecasted to be more Republican-leaning than the nation.

“Forecasted” is an important word there. These states voted to the right of the country in 2016 — when, of course, Hillary Clinton lost the Electoral College despite winning the popular vote — and in 2020. And the model expects that to happen again in November.

I’ll grant you that there’s a little bit of ambiguity in what’s happening in the polls right now. Since the debate, Biden has registered a slightly steeper drop in our national polling average than in our polling averages of these swing states, but that’s partly a result of a comparative lack of high-quality state polls.² By contrast, our forecast for November blends the current polling averages with estimates based on “fundamentals” and regression analysis, which tends to smooth out some of the kinks in the data. It’s the forecast that I’ll be talking about for the rest of this newsletter.

The table below shows the states the model thinks are most likely to put the winning candidate over 270 electoral votes. These nine states decide the election in about 91 percent of our simulations, and 7 of the 9 are expected to vote more Republican than the country overall. Only Maine and Michigan — which collectively tip the election in 12 percent of our simulations — are projected to vote to the left of the country.³