The worst-case scenario for the Dems in North Carolina is one of the lower Black turnout (by the roughly 7 points’ worth of the electorate we see in the early voting now) continuing in the next few days and a White vote more Republican-voting than that in 2012, with unaffiliateds staying just as Republican as in 2012, registered Republicans staying just as large a percentage of the White vote as in 2012, and registered Dems becoming a much lower percentage of the White vote. In that case, Trump wins North Carolina by 8 points. Registered Republicans growing to as large a percentage of the White vote as in 2012 is plausible; unaffiliateds staying just as Republican-voting as in 2012 isn’t.

The most plausible scenario we see on November 9 (accounting for the FBI reopening its investigation, Trump likely gaining support in polls from now until November 8, etc.) is one of lower Black turnout (by approximately 5 points’ worth of the electorate) and a White vote somewhat more Democratic-voting than in 2012 (though not as Democratic-voting as that seen in the early vote now), leading to a four-point Trump win in NC.

The scenario seen now in the early voting is one of lower Black turnout (by 7-8 points of the electorate) and a nearly equally Democratic-trending White vote to offset it. If this remains the pattern of the early vote relative to 2012’s from now until election day, with no Trump polling improvements whatsoever, Trump wins North Carolina by 2 or 3 points. Romney won by 2.

The worst-case scenario for Trump is Black turnout returning to 2012 levels and the White vote staying as Democratic-leaning as it is in the early vote now or becoming more Democratic-leaning. In that case, Trump loses by four points.

In none of these circumstances does the Upshot scenario of a Trump six-point loss in North Carolina appear even remotely plausible. A six-point Trump win under realistic assumptions is certainly plausible. A six-point loss isn’t. The Upshot’s assumptions about the unaffiliated vote are not even remotely realistic.

It seems that the NC voters who voted this year, but not in both 2012 and 2008, are, even adjusted for higher Black turnout in those years, slightly less-Democratic-registered than those who voted both in 2008 and 2012. But, overall, it’s a wash with the new non-Black voters relative to the old ones, and Black turnout is the biggest driver of Republican success relative to 2012 so far in NC.

If North Carolina is representative of the entire country, Trump is much more likely to win Florida than FiveThirtyEight says, but the Black vote is not nearly as important in other swing states.