A nice round-up by CNN of indicators pointing to a Republican win in NC, although all early-voting projections come with a big asterisk. Namely, how will whites, especially those with a college degree, vote? For all the hype about Trump’s strength with white voters this year, he actually underperforms Romney with that group in many polls. Case in point: Romney won whites nationally by 20 in 2012 but Trump led them by just 17 in the final Fox News poll. If Hillary holds more or less steady in black support from 2012 and gets a boost in Latino votes and Trump can’t offset that by outpacing Romney with whites, well, you don’t need an electoral model to tell you how this is going to go.

As of Saturday, the final day of early voting [in North Carolina], slightly fewer Democrats had cast ballots [compared to 2012] while 125,000 more Republicans have voted this time. If this election shapes up like the last, Donald Trump would win North Carolina.

There is one key difference that complicates the data: Independent voters came out this time in droves. They cast nearly 810,000 votes, up a whopping 42% from 2012…

Other early voting trends favor Trump. While polls show Clinton has a commanding lead among African Americans, the share of black votes so far is down 5 percentage points. Trump does much better with white voters, who increased their share by 22% this election. The North Carolina Republican Party, in a statement Monday, bragged about the changes in turnout as a sign of the “North Carolina Obama coalition crumbling.”…

One third of Democrats age 22 to 29 who voted in 2012 failed to show up this time… Twenty-something Republicans were more enthusiastic. Nearly three quarters of them who voted early in 2012 showed up again this election.

Democrats topped Republicans by more than 300,000 early votes in North Carolina but that’s a smaller margin than Dems enjoyed in 2012, when Obama lost the state narrowly to Romney after strong Republican turnout on Election Day. The questions now are whether black voters who skipped early voting this time will show up after all, and of course whether the white votes that pile up will split decisively enough for Trump to erode Hillary’s early-vote lead. The last Upshot poll of the state found Trump winning whites with 59 percent of the vote, which is strong — but not as strong as Romney, who took 68 percent of the white vote in the state en route to a very narrow victory four years ago. See why North Carolina is so confounding for modelers? Not only is it one of the tightest states in the country, there are signs that each nominee could underperform among the groups they’re counting on. Hillary’s black early vote is down, but Trump is weaker among whites (thanks to college grads) than the last GOP nominee was. Where does that get you prediction-wise?

And where does this fit in the larger scheme of early voting results? North Carolina is must-win for Trump, but then pretty much every battleground is must-win. The must-must-must-win is Florida, without which he’d have to bump off multiple Rust Belt states that have been reliably blue in recent years to have any chance. The best summary of Florida early voting I’ve seen comes from Democrat Steve Schale, who worked for Obama and Clinton. Discount his opinion if you like accordingly, but I don’t know what incentive he’d have to lie at this point about what he sees happening in the state. The race is tight and white turnout is up, but the electorate nonetheless looks to be more diverse overall than it was in 2012 — which is good news for Hillary:

But I want to start with a couple of numbers. First: 67. 67% was the percentage of the electorate was white in 2012 — which by the way was down from 71 in 2008. My foundational assumption was if the electorate was more diverse than 2012, the basic coalition that got President Obama over the line in 2012 would hold. We finish early voting at 65.7 white, 15.3 hispanic, and 13.1 black, with the black number closing in on the 2012 share, and the white number down…

While Republicans talked about Trump’s ability to turn-out low propensity voters, it is Clinton who has turned out 250,000 more low propensity voters.

[Unaffiliated] voters, making up the largest share they’ve ever made up in a Florida Presidential election, are 4 points more diverse than the electoate at-large, including a 20% Hispanic share…

[F]or Trump to win, he has to have a ridiculously good day. I suspect that when early voting is counted, that she will have won the early vote by 3-4 points, and if early voting is, let’s say 2/3rds of all the votes, it means Trump has to win [on Election Day] by 6-8 points. I don’t think 6-8 points is out there today for him.

Schale says that, on a scale of one to 10, he’s at a nine confidence-wise, which is more confident than he was before Obama’s narrow win in Florida four years ago. He’s expecting a Clinton victory in the range of 1.5 to 2.5 percent. Needless to say, though, with a margin that narrow, even a small underestimation in Trump’s share of support among whites (or among Latinos) could tilt the state.

I’ve been amazed to see so many data nerds over the last 24 hours predicting Clinton will end up with 322 electoral votes, which assumes victories in Florida and North Carolina notwithstanding how close the two states are and the fact that white turnout was up in early voting across multiple battleground states, a good sign for Trump’s national prospects. In the end, I guess, they’re counting on Hillary’s ground-game advantage to make the difference. Read this post by David Drucker quoting multiple Republican “insiders” as worrying that, in the end, the RNC simply can’t keep up with the Clinton/Democratic operation. That’s mostly Trump’s fault: By eschewing his own GOTV effort and outsourcing traditional campaign turnout tasks to the RNC, he assured a smaller, overtaxed Republican operation and one that would be less coordinated in translating data into strategic choices. (A few days ago, the NYT reported that some of Trump’s aides were using the website 270ToWin to demonstrate his paths to victory, which is sort of like a general explaining his battle plan using the board game Risk.) But Trump’s not entirely to blame. One GOP analyst interviewed by Drucker said this of the RNC’s operation: “The reality is that Trump has less than nothing and the RNC has a framework of 2012 filled in with more smoke and mirrors than a haunted house… We’re going to find out Tuesday that literally everything the RNC has been selling was focused on improving the perception of their efforts rather than the product.” Gulp.