This new one from CBS/YouGov is actually three polls in one. You’ve got your North Carolina and Pennsylvania data, sure, but there’s also a mega-poll of 13 different battleground states that has the race surprisingly close: Clinton 42, Trump 40, well within the margin of error. That’s the good news for Trump fans here — in the states he really needs to win, he’s right there with Hillary as we move to the home stretch of the campaign.

The bad news is that he’s down four in North Carolina and eight in Pennsylvania, and both of those states are effectively must-win given the reality of the electoral map. In fact, if Hillary locks down PA and New Hampshire (where she leads by nine points or better in recent polls) and hangs on in Wisconsin, where the race has narrowed but which has gone reliably blue in recent decades, that would put her at 259 electoral votes with all sorts of paths still on the map to 270. Any one state among Ohio, Florida, North Carolina, Virginia, Georgia, or Arizona would get her there. She could even get there by pulling an upset in Missouri and winning either Maine’s or Nebraska’s second congressional district, each of which awards a single electoral vote independent of the state result. She doesn’t need North Carolina. Trump does, plus a whole lot more.

This week, Clinton leads in two new polls of states that Trump needs to win, given his current electoral map: Clinton leads in Pennsylvania fairly comfortably by eight points, 45 percent to 37 percent, and she is up four in North Carolina, 46 – 42 percent…

For Clinton, doubts about her explanations of the email server continue to weigh on her. Forty-six percent across the battlegrounds say those explanations are changing and getting less believable. The same percentage say those explanations remain the same. Only 7 percent feel the explanations are getting more believable…

In a week that saw Donald Trump’s trip to Mexico and speech on immigration, more voters (47 percent) across the battlegrounds feel his policy on immigration has not changed, while 37 percent say he has changed policies to become easier on those in the U.S. illegally. Neither of those views appears to have done much to his vote share, though it illustrates the difficulty he may have satisfying both his base and trying to broaden his appeal on the topic.

Her average lead in the four-way race in Pennsylvania right now is a solid six points and would be higher if not for the questionable landline-only survey by Emerson College last week that put her up by just three. (Polls that don’t include cellphones tend to underestimate Democratic support.) Trump has picked up a few points since his post-convention swoon in PA, when he was down 9-10, but the state still doesn’t look particularly competitive. In 12 four-way polls taken there since July 1st, he’s cracked 42 percent or better only twice. Hillary has fallen to 43 percent or worse only twice in the same 12 polls.

No one wants to read a separate analysis of each of these three polls so let me flag a few interesting results from each:

1. One question that many pollsters have asked this summer is whether supporters of each candidate are voting mainly for them or against the other party’s nominee. Typically Hillary does better than Trump on that metric: Clinton voters tend to say they’re voting for her whereas Trump voters tend to say they’re voting against Hillary. YouGov refined the question, though, to give people a choice in saying whether they’re voting for their candidate because they like him/her personally or because he/she is the party’s nominee. In each case, true to form, roughly half of Trump’s voters say they’re voting to stop Clinton while the number of Clinton voters who say they’re voting to stop Trump ranges from 36-43 percent. His base continues to be more anti-Hillary than pro-Trump, which is what you’d expect given that his politics aren’t agreeable to some Republicans.

What’s noteworthy, though, is the split between those who say they’re voting for their candidate because of personal appeal versus out of partisan duty. In two polls, Trump tops Clinton among supporters who say they’re voting the way they are due to their candidate’s personal appeal; Hillary, however, leads him by double digits in all three polls among those who say they’re supporting their candidate purely because he/she is the nominee. Maybe Democratic partisans are more loyal than Republican ones, which wouldn’t be surprising in light of how many GOPers have come to disdain the party. Or maybe it’s a product of Trump drawing more independents. Naturally someone who doesn’t identify with either party won’t vote for Trump out of partisan duty.

2. YouGov asked people who aren’t voting for Trump or Clinton right now whether there’s a chance they might reconsider by November. Slightly more people in Pennsylvania and in the 13-state poll said they would never consider voting for Clinton than said so for Trump. But don’t celebrate just yet. When you multiply the number in PA who aren’t voting Trump right now by the number who say they’ll never vote for him, you get 52 percent of the overall electorate who are firmly opposed to him. For Clinton, by comparison, it’s just 48 percent. Assuming those people are telling the truth, Trump’s only hope of winning the state is Gary Johnson picking off tons of Hillary voters. The numbers are the same in North Carolina too: 52 percent of the electorate claims it’s #NeverTrump while 48 percent claims it’s #NeverHillary. If those numbers don’t change, a strong Gary Johnson performance at Hillary’s expense is Trump’s only hope of victory.

3. Amazingly, although Trump trails in all three polls, his numbers on each of these individual questions are better than Hillary’s in each survey — and in some cases significantly better.

That’s from the 13-state battleground poll but the trend is the same in the NC and PA data too. More people say that Hillary’s campaign, not Trump’s, is at least somewhat based on appealing to certain racial groups over others. Many more people in each poll say that Hillary’s personal desire to be president is a strong factor in her candidacy, while more say Trump has a sincere desire for a better America. And on the question of which of them will do what party leaders and donors want, it’s no contest. Clinton is at 85-86 percent(!) in all three polls among people who say her campaign is based “a lot” or “some” on serving establishment masters. Trump tops out at 36 percent. There’s a 50-point difference. Respondents also give Clinton toxic numbers when asked if the phrase “tells the truth” describes Clinton. Her best mark there is 33 percent in North Carolina. In Pennsylvania and in the 13-state poll, it’s 28 percent. How is this person leading?

The answer, I think, lies partly with what happens when people are asked if Trump and Clinton are each prepared to be commander-in-chief. In each poll, Clinton cracks 50 percent or better while Trump falters at 35-37 percent, with 63-65 percent saying that phrase doesn’t describe him. That’s been true all campaign long, across many polls. The public knows Hillary is a lying sleaze, but they think the sleaze can be trusted with nuclear weapons. They continue to doubt that about Trump. He’ll get three shots at convincing them otherwise in the debates this month and next. If he can’t, he’s likely done, no matter what new Clinton dirt emerges. In fact, a footnote on the debates: Across these three polls, seven to 10 percent more people say they’re aware of where Clinton stands on the issues than where Trump stands. The debates will be more of a policy introduction for him to the audience than they will be for her. He has more room to grow than she does, in theory.

4. Interestingly, Trump is polling a respectable-for-a-Republican mark of 30 percent among Hispanics in the 13-state poll while Hillary tops out at 55 percent. In 2012, Obama topped Romney nationally among Latinos by a 71/27 margin. Trump is well ahead of that pace in the battlegrounds. Is that proof that he’s going to outperform Romney nationally with Latino voters this year, despite the endless doomsaying of the pundit class that his immigration stances are poisonous within that group? Not necessarily: It could be that Trump will get utterly swamped among Latinos in solid blue states like California, which will pull his national average towards being more in line with Romney’s. I.e. maybe Romney was somewhat more competitive in 2012 with Latinos in the 13 battleground states YouGov polled this week than he was nationally. A quick check of the exit polls, though, shows that he lost Latinos 80/18 in Pennsylvania that year and 68/31 in North Carolina, both of which are considerably worse than Trump’s 13-state average right now. It may be that Trump really is quietly overperforming Romney in key states — although he’d probably need to overperform by a lot for it to make a difference. Latinos were just six percent of the Pennsylvania electorate in 2012 and just four percent of the North Carolina result. Even a major shift within that group is probably a matter of less than a single percentage point overall.

5. It’s taken for granted among Trumpers that their side is more motivated to vote for Trump than Democrats are for the charmless Clinton. But is it true? Here’s a tidbit from the 13-state poll on voter enthusiasm:

Interestingly, those numbers are starkly different in Pennsylvania. There, 36 percent of Democrats say they’re enthusiastic to vote for their candidate while 41 percent say they’re satisfied. For Republicans, it’s 47 and 29 percent, respectively. And yet Pennsylvania is Trump’s worst poll of the three YouGov conducted. Why is that? The answer, I think, is that his enthusiasm numbers look higher in PA because it’s mostly true enthusiasts who are voting for him there. He’s not getting the share of “satisfied but not enthusiastic” voters that a nominee would normally expect. If you doubt that, check out the numbers in Pennsylvania when people are asked whether they’re more motivated to vote this year or less:

Just 13 percent of Dems say they’re less motivated to vote this year. Republicans? 22 percent. That’s the biggest gap in the three polls. There’s actually no gap at all in North Carolina — it’s 14 percent for both parties — and in the 13-state poll, the gap is five points. There’s some serious Republican demoralization happening in PA, which is probably why he’s stuck well below 40 percent. Which brings us to…

6. Right now the biggest problem for Trump in Pennsylvania and in the 13-state poll is how poorly he’s performing among Republicans. Check out the partisan gap in PA:

Hillary’s at 83 percent with Democrats. Trump, among Republicans, is at just 75. Among self-described conservatives he’s polling even worse, at 73 percent. Same story in the 13-state poll:

His raw number with Republicans is higher here than in Pennsylvania, but then so is Hillary’s. In fact, at nine points, the partisan gap between them is wider here than it is in Pennsylvania. That’s not a huge deal in PA since her lead is so large: Even if he were winning 83 percent of Republicans there, it would probably put him at only 40 percent or so overall, still several points behind Clinton. It’s a bigger deal in the 13-state poll, though, since Hillary’s lead there is just two percent. If Republicans were lining up for Trump the way Democrats are for Hillary, he’d probably be leading narrowly overall.

Is that proof that #NeverTrumpers have sunk the Trump? No, starting with the fact that many of the Republicans who say they’re not voting for him right now haven’t actually ruled out voting for him. in Pennsylvania, four percent of Republicans who aren’t backing him say they’d consider backing him in November and another 40 percent say they might consider it. Among conservatives those numbers are four and 35 percent, respectively. They’re not #NeverTrump, they just haven’t been sold on him yet. Beyond that, given how Trump tends to underperform in various polls among college-educated whites, white women, and suburbanites — all of whom are traditionally Republican voters — my guess is that his problem with anti-Trump Republicans and conservatives is less an artifact of strong ideological objections to Trump than it is of class- and gender-based objections. These people aren’t resisting Trump because they read a hot Jonah Goldberg column that changed their minds, they’re resisting him in all likelihood because the way he talks about issues with strong class components, like immigration, puts them off somehow. And ironically, I think Team Trump itself agrees. That’s why he went down to Mexico to meet the president and why he’s visiting black communities in Detroit this weekend. He’s not worried about conservatives thinking he’s too moderate for them to support, he’s worried about suburban Republican-voting moms thinking he’s not moderate enough.

In lieu of an exit question, here’s pollster Kristen Soltis Anderson weighing in on the possibility of “undercover Trump voters” delivering him the election.

Pollsters @KSoltisAnderson and @MargieOmero: Theories of "hidden Trump vote" overblown, "a sign of real weakness." https://t.co/wyy4jQGeW9

— This Week (@ThisWeekABC) September 4, 2016