He hasn’t trailed in a poll of Georgia since early August and that’s still true today, but with this new one from Quinnipiac showing him up just 44/43, his lead in the RCP average has shrunk to 2.8 points. In Iowa, where Quinnipiac has it 43/43, his lead in the average is down to 1.4 points. That’s supposed to be one of his best swing states because the population is overwhelmingly white and comparatively few residents have college degrees, which is his base. Neither one of these states can go red if he’s hoping for a shot at 270, needless to say.

Although Quinnipiac polled four states — Clinton leads in North Carolina and Virginia too — it’s the Georgia number that’s getting the most attention. Partly that’s because Georgia isn’t supposed to be a Republican stronghold and partly it’s because of this:

Clinton leads 48 – 42 percent among Georgia early voters.

Georgia likely voters have negative opinions of both Clinton and Trump, with Clinton at a negative 40 – 55 percent favorability rating and Trump at a negative 38 – 55 percent.

In the four-way race, men back Trump 49 – 39 percent, as women back Clinton 47 – 39 percent. White voters back Trump 65 – 23 percent, while non-white voters back Clinton 77 – 9 percent.

Trump takes Republicans 86 – 7 percent while Democrats go to Clinton 93 – 3 percent.

Romney won the early vote in Georgia in 2012 by four points. That doesn’t mean Trump can’t hold the state, but it’s evidence that Democratic turnout is solid in the first days of voting. Interestingly, Georgia doesn’t show the educational divide this year that many states do in which Clinton leads among college-educated whites while Trump leads big among whites without a degree. Quinnipiac shows him leading whites in GA across the educational spectrum, by 27 among college grads and by 53(!) — nearly double the margin — among the working class. (Which makes sense, since the white majority there is solidly Republican.) The reason the race is close anyway is because (a) Clinton is crushing him among the state’s nonwhite (i.e. black) minority, 77/9, and (b) she leads by eight points among women despite Georgia usually being reliably Republican. Nationally, a best-case scenario for Trump this year would have looked something like this poll, piling up large majorities among whites of all classes to overcome Clinton’s enormously large majorities of nonwhites. As it is, Georgia is competitive because Trump, despite holding a sizable lead among white college grads, hasn’t made it quite sizable enough. I think that’s what you’re seeing in his weaker support among Republicans (86 percent) relative to Clinton’s stronger support (93 percent) among Dems. The number of Republican-leaning white college grads tilting towards her in Georgia is smaller than it is in most battlegrounds, but still large enough to make a difference. And the number of women he’s losing to Clinton is eroding most of his advantage with men.

Obama won Iowa, remember, so the fact that it’s tied is a positive sign for Trump — sort of, since its demographics suggest he’d play better there than Romney. But his momentum is pointed the wrong way. A month ago he led a Quinnipiac poll of the state by seven, with a 26-point lead among men versus just a 10-point lead for Clinton among women. Today Clinton leads women in Iowa by 15 while Trump’s lead among men has slipped to 16. Among early voters she’s up, er, 61/27. In mid-October of 2012, Dems led Republicans in early ballot requests in Iowa by 15 points.

As for North Carolina and Virginia, he’s down 12 in the latter in this poll but that doesn’t much matter since VA hasn’t been very competitive all year. With one exception, Clinton has led by five or more in every poll over the last month. Few forecasters include it in Trump’s hypothetical paths to 270 — and yet Trump held a rally there just five days ago. Don’t ask me why. North Carolina is tighter, with Clinton ahead by four in this Quinnipiac poll but just 2.4 in the RCP average. Note the small but significant divergence in support for the nominee within each party:

Romney won Republicans in North Carolina in 2012 by a 96/4 margin while Obama won Democrats 91/8. Clinton’s improving on Obama’s performance this year while Trump is nearly 10 points off the pace of Mitt’s. How come? It’s probably due to women, including some Republican women bailing on Trump. Women split very narrowly for Obama in North Carolina four years ago, 51/49. Clinton’s winning them here by 15 points. And although Quinnipiac doesn’t give crosstabs on college grads of all races (just white college grads), the fact that white college grads are nearly even in this poll suggests that she’s way outperforming Obama among that group too. In 2012, voters of all races with college degrees in NC split 50/49 for Obama, probably because nonwhite college grads broke very heavily Democratic while white college grads broke solidly Republican. This year the latter group is itself narrowly divided. That’s how you lose North Carolina. And … maybe Texas? A new poll out today from the Texas Tribune finds Trump’s lead there down to three because, again, Clinton commands more loyalty from Dems (93 percent support her) than he does from Republicans (83 percent support him). Again, that’s probably a function of defections among women, who are split 45/45 in Texas. In 2008, when he got blown out nationally, McCain won women in Texas by five points.

Exit question: Can some belated ObamaCare messaging help with this? White college graduates are, I’d guess, a group that’s disproportionately affected by soaring premiums, since many may be sufficiently well off that they don’t qualify for subsidies. Maybe this will bring some back into the fold.