I wondered this morning why there aren’t any recent polls from Florida and Ohio and this afternoon I’m getting not one but two from each for my trouble. One will be published by NBC/WSJ at 5 p.m. ET (about an hour from now as I write this). The other is this new one from Quinnipiac, which has some reassuring numbers for Trumpers after a tough 10 days. The good news: He’s still in the thick of it in Florida and Ohio, down just two points in the latter in a four-way race and dead even in the former. Those are two of the three states he needs to have a shot at 270 electoral votes this fall. He’s right there.
It’s the third part of the trifecta that’s causing him trouble. Pennsylvania: Clinton 52, Trump 42. In a four-way race, Clinton leads by nine.
Why focus on the Pennsylvania numbers instead of the better Florida and Ohio ones? Again, because he needs all three states to bank the electoral votes he’ll need to get to 270. It’s technically possible, of course, that he could lose PA and make up the votes elsewhere by winning other purple states, but true purple states are scarce now. Virginia and Colorado, which used to be battlegrounds, are pretty safely blue in 2016. Trump could try to go get Maine, New Hampshire, and one or two others, but it’d be odd for him to lose Pennsylvania yet see a series of lesser targets come through. If the PA electorate is strongly pro-Clinton, odds are swing voters in other swing states are too.
As for the raw numbers in the table above, you know the story by now. Trump wins men — but not as heavily as Hillary wins women. Trump wins a few Democrats — but not as heavily as Hillary wins Republicans. Trump wins whites without a college degree easily — but Hillary wins whites with a college degree, cutting deeply into the wide margin he needs among white voters overall to offset her strength among nonwhites. In fact, as you can see above, Hillary actually wins white women in Pennsylvania by fully 10 points, not nearly as much as Trump’s margin among white men but more than enough combined with her strength among minorities to give her a comfortable lead.
Compare those numbers to the numbers in Florida, where it’s dead even:
Trump has much stronger support among whites in Florida, winning both the college-educated and white women. Hillary’s gender gap is smaller too, winning women by 13 points instead of the 22-point gap she enjoys in Pennsylvania. Her advantage among nonwhites in Florida is weaker, from 79/8 in PA to 66/20 in Florida. (I assume that’s a function of many Cuban-American voters in Florida leaning Republican.) Trump even tops Hillary in the partisan battle, winning Republicans 90/5 versus Hillary’s 86/7 share of Democrats. One footnote to all of this, though: After Nate Silver tweeted this afternoon that the numbers here in Florida and Ohio were about what he expected, I replied that they seemed surprisingly close to me given the beating Trump’s been taking in national polls lately. Trump in a dead heat in Florida, a must-win state for him, is encouraging news. Silver answered that once you account for Quinnipiac’s “house effect,” which tends to favor Republicans by three points or so, Florida is more likely a three-point lead for Clinton, Ohio a five-point lead, and Pennsylvania a — gulp — 12-point lead. The “house effect” is something that his site, FiveThirtyEight, derives by comparing a pollster’s numbers over time to numbers other pollsters get for the same races. If a pollster reliably sees stronger numbers for one party or the other in poll after poll, race after race, it’s presumed to have a partisan lean built into its model of the electorate. It’s a form of unskewing, ironically enough, although it’s based on statistical measurements of many different elections, not just a guess at what the electorate “should” look like this fall. If you’re taking the house effect into account here, Clinton leads narrowly in Florida, fairly comfortably in Ohio, and is blowing Trump out in Pennsylvania. You can see Silver’s adjustments for each poll his site tracks by comparing the “leader” to the “adjusted leader” column in each state.
Anyway, that’s Quinnipiac. Now we wait for WSJ/NBC. While we do, via the Right Scoop, there are new national numbers from Reuters’s daily tracker. That’s the poll that caused a stir this past weekend when it found Trump within 2.4 points of Clinton at a moment when various other polls had her leading somewhere between seven and 10. Today’s data: Clinton 42, Trump 35. Clinton by seven, right in line with her national average.
Update: And here are those WSJ/NBC numbers now. My mistake: They didn’t poll Florida. The swing state they’re focused on is Iowa.
New NBC/WSJ/Marist polls:
IA: Clinton 41, Trump 37 (+4)
OH: Clinton 43, Trump 38 (+5)
PA: Clinton 48, Trump 37 (+11)
— Mark Murray (@mmurraypolitics) August 9, 2016
Those Ohio numbers are right in line with FiveThirtyEight’s “adjusted lead” for Clinton in the Quinnipiac survey. But never mind that. The news, obviously, is Pennsylvania, where Clinton is once again leading by double digits. That makes four polls of the state in which she’s ahead by 10-11 points; Romney lost there by five but Trump, because of his working-class appeal, was supposed to be more competitive. If Pennsylvania’s a goner, it’s not an exaggeration to say that the election is all but over. Trump will have to hold every last one of Romney’s states, win Florida and Ohio, and still find another 17 electoral votes among Obama states to pry loose. Where do those votes come from if not from PA?
By the way: His maximum vote share in three key states is 38 percent?
Update: Here’s what Trump needs to pull off if Pennsylvania is off the board.
If you take PA off the map for Trump, he has to win EVERY ONE of FL/IA/NV/NH/NC/OH — if you give HRC CO/VA pic.twitter.com/3ZNpFsXTUK
— Mark Murray (@mmurraypolitics) August 9, 2016
The odds of him winning all of those states/districts while also getting blown out in PA are vanishingly small. He needs to make something happen in Pennsylvania.