Final election prediction: Biden projected to win by 7 million votes - but is that enough?

On Election Eve, there’s a better sense of what the actual outcome is going to be than in any US presidential race in history. The simple reason: There’s more of the vote already in than ever before.

As of Monday morning, 100 million Americans have already cast their vote — that’s about 42 million more than in 2016 and about 73 per cent of the total votes cast for president in 2016.  CBS News/YouGov has polled those people and found that Joe Biden leads with these voters 66 per cent to 32 per cent. But included in the poll was a sample of the people who said they were voting on election day and that favored President Trump, 69-to-27 per cent over Biden.

Most are expecting about 150 million Americans to ultimately cast their vote. Nate Silver of FiveThirtyEight more precisely predicts 154 million voters — or about 17 million more than last cycle. Using these numbers and the CBS News/YouGov poll, we can forecast the final vote totals. It’s simply 66 per cent of 100 million (66 million) plus 27 percent of the remaining 54 million (14.6 million) for Biden for an expected 80.6 million Biden votes. That’s compared with about 32 million early votes for Trump (32 per cent of 100 million)  plus 37.3 million predicted election day votes (69 per cent of 54 million), for a Trump total of 69.3 million votes. So 80.6 to 69.3 translates into 52.3 per cent for Biden and 45 per cent for Trump, which reflects the average current combined polling averages of 6.5 points at RealClearPolitics, which does not weight polls for quality and sample size, and the  8.5 points at FiveThirtyEight, which does.

Given that Trump lost the last election to Hillary Clinton by 2.1 points yet still managed to prevail at the state-level in the Electoral College, could Trump overcome a national vote margin remotely approaching this?

That’s very unlikely. A Vanderbilt University study concluded, “Based on thousands of simulations, …the bias in 2020 probably will favor Trump again but to a lesser degree than in 2016.” According to the study, Trump’s chance of winning with even 48 per cent of the vote is seen as “remote,” about 12 per cent. That’s just slightly more than the 10 per cent chance the FiveThirtyEight model gives him as of Monday.

And of course a major caveat in the CBS News/YouGov poll is that it’s much more accurate to find out how people have actually voted than how they hypothetically say they are voting, given those people may not vote at all or could change their minds. So the confidence level on the early vote polling should be much greater than the hypothetical share that either Biden or Trump will get in votes to come.

Another potential wildcard in Biden’s favor is that the minority vote has not turned out to the extent expected in the early voting. But these voters are more likely to lack trust in early voting and thus are more predisposed to voting on election day. If this is true, the demographics of the election day vote will be significantly different, meaning likely more favorable to Biden, than the pollsters presently expect.